Mitochondrial creatine kinase 1 regulates the cell cycle in non-small cell lung cancer via activation of cyclin-dependent kinase 4

The human NSCLC cell lines A549, PC9, and H1299 were preserved in the State Key Laboratory of Oncology in South China, and cultured according to their guidelines. A549, PC9, H1299, and the corresponding modified cells were cultured at 37 °C in RPMI1640 (Gibco, USA) supplemented with 10% fetal bovine serum (Invitrogen, USA), 100 U/mL penicillin, and 100 μg/mL streptomycin in a humidified atmosphere containing 5% CO2. HEK-293 T cells were cultured in Dulbecco’s modified Eagle medium (Gibco, USA). All cell lines were authenticated by short tandem-repeat (STR) DNA profiling (Microread Diagnostics Co., Ltd, Guangzhou, China).

Complementary DNAs of CKMT1 and CDK4 were inserted into the pCDH-EF1-MCS-T2A-Puro vector with 3 × Flag at the N-terminus (System Bioscience, Palo Alto, CA, USA), and short hairpin RNAs (shRNAs) targeting CKMT1 (CKMT1-sh1: CGTGGAATTTGGCACAACAAT and CKMT1-sh2: CGGTGTCTTTGATATTTCTAA) and negative control shRNA (shNC: CAACAAGATGAAGAGCACCAA) were cloned into a pLKO.1 vector (Sigma-Aldrich, St. Louis, MO, USA). The doxycycline-inducible lung cancer cell lines were constructed using Tet-on inducible expression system following the manufacturer’s instructions. These plasmids were then packaged in 293 T cells to obtain recombinant lentiviruses with a Lentiviral Packaging Kit (FulenGen). Following a 48 h period of infection with lentivirus plus 5 mg/ml Polybrene, stably overexpression or knockdown cell lines were selected with 3 μg/mL puromycin for 3 days as previously described [20, 21].

We transfected small interfering RNA-targeting candidate genes or Negative control (NC) siRNA (Sigma) into adherent cells using Lipofectamine 2000 or 3000 reagent (Invitrogen), according to the manufacturer’s guidelines. The sequences of each siRNA are listed in Additional file 2: Table S1.

CKMT1 knockout cells were generated using the CRISPR-Cas9 system. The single-guide RNAs (sgRNAs) sequences targeting CKMT1 were designed from Zhang’s laboratory website (http://​crispr.​mit.​edu/​). The sgRNA sequences were sgRNA1: TACGAGCTGCCAGTGAACGA and sgRNA2: GGACCGACTAGGCAAATCAG. The sgRNAs were cloned into the Lenti-CRISPER v2 vector (Addgene#52961). Lentiviral particles were produced in 293 T cells. Positive clones were selected with 1 μg/ml puromycin and confirmed by DNA sequencing. The loss of CKMT1 protein expression was verified with the CKMT1 antibody by Western blot.

Immunoprecipitation (IP) and Western blotting (WB) were performed as previously described [22‐24]. Briefly, cultured cells were lysed in lysis buffer (CST, MA, USA) with PMSF, protease inhibitors (Roche, Basel, Switzerland) and phosphatase inhibitors (KeyGen Biotech, Nanjing, China). Protein concentrations were measured with a BCA protein assay kit (Beyotime, Haimen, China) according to the instruction. For IP [22], cell lysates were incubated with unconjugated primary antibodies at 4 °C overnight, followed by 4 h incubation with protein G agarose beads (Santa Cruz, CA, USA) at 4 °C. Then the precipitated protein complexes were washed four times with Tris-buffered saline (TBS) containing 0.05% Tween (TBS-T), boiled in 1 × loading buffer for 10 min, and then analyzed by WB. For WB [23, 24], protein samples were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene fluoride (PVDF) membranes. After blocking nonspecific binding in 5% non-fat milk, the blots were incubated with specific primary antibodies against CKMT1, CDK4, p-CDK4, c-Myc, IgG, Flag, VDAC1, p-Rb, β-Actin, β-Tubulin, COX IV, GAPDH and GST, followed by reaction with HRP-conjugated antibodies (CST, MA, USA). All antibodies were diluted to 1: 1000 in the blotting. Signals were visualized using an ECL detection system (Amersham Biosciences).

Purification of recombinant GST fusion proteins expressed in E. coli Agilent BL21 was transformed with pGEX4T1 or its derivatives containing the CKMT1, CDK4 coding sequences. GST pull-down reactions were conducted as follows [25]. Beads (50 μl) were coated with 10 μg GST fusion protein or GST control in PBS and incubated with specified amounts of soluble recombinant protein or cell lysates in a final volume of 0.5 ml of reaction buffer (50 mM HEPES, pH 7.4, 5 mM MgCl2, 150 mM NaCl, 0.1% Triton X-100, 0.1% bovine serum albumin (BSA) and 1 mM PMSF). The samples were incubated at 4 °C for 6 h or overnight while rotating. The beads were washed three times with PBS at 4 °C, boiled in 1 × loading buffer for 10 min, and then analyzed by SDS-PAGE.

Cells in confocal dishes were fixed in 4% paraformaldehyde, permeabilized using 0.5% Triton X-100, and blocked with 3% BSA (Beyotime, Shanghai, China). After being blocked in BSA for 1 h, the cells were incubated with primary antibody against c-Myc or Flag or CKMT1 or CDK4 at 4 °C overnight. Next, the secondary antibody conjugated with Alexa Fluor 488 Dye (Invitrogen, CA, USA) or Alexa Fluor 594 was used to label the primary antibody for 1 h at room temperature in the dark. Then the samples were co-stained with DAPI (Beyotime, Shanghai, China). Photographs were captured and analyzed using a confocal laser scanning microscope.

The Dataset GSE19804 was obtained from NCBI Gene Expression Omnibus (GEO) (https://​www.​ncbi.​nlm.​nih.​gov/​geo/​) and used to analyze expression differences of kinase genes in NSCLC and control lung tissues. Total RNAs were extracted from cultured CKMT1 knockout A549 cells and controls (or cells stably expressing shCKMT1 and control vector) using TRIZOL Reagent (Invitrogen, CA, USA) according to the manufacturer's instruction. The quantity and quality of total RNAs were measured and purified. Next, the total RNAs were subjected to Affymetrix PrimeView Human Gene Expression Array conducted by Boho Biotech Company, Shanghai, China. Differentially expressed genes (DEGs) between the CKMT1 knockout A549 cells and control cells were analyzed by gene set enrichment analysis (GSEA) software (Broad Institute, San Diego, USA) to find gene sets enriched by CKMT1 knockout or knockdown. Pathway enrichment analysis was performed using KEGG.

Cell proliferation was assessed by CCK-8 Cell Counting Kit (Dojindo Laboratory, Kyushu, Japan) and colony formation assay. For the CCK-8 assay, cells were cultured into 96-well plates at a density of 2000 cells per well and incubated for 7 days. The growth rate of cells was determined by OD_450nm using CCK-8. For colony formation assay, cells (1000/well) were seeded into 6-well plates and cultured for 10–14 days, followed by staining with 0.5% crystal violet. Images of Colonies were captured and the number of colonies was counted by Image J software.

Cell cycle assay was performed by flow cytometry. Cells were fixed with ethanol 70% and incubated at − 20 °C for at least 24 h. Cells suspension was subsequently centrifuged and washed with ice-cold PBS, followed by staining in propidium iodide (PI) (20 μg/mL RNase, 50 μg/mL PI, and 0.1% (v/v) Triton X-100 in PBS) for 30 min at 37 °C water bath in the dark. The cell cycle stage was determined by flow cytometry and analyzed by FlowJo software.

Animal experiments were approved by the Sun Yat-sen University Cancer Center Institutional Animal Care and Usage Committee. Female BALB/c nude mice (4–5 weeks old, 15–18 g) were purchased from the SLRC laboratory Animal Co. (Shanghai, China). Mice were used to generate xenograft models via subcutaneous transplant of tumor sections (approximately 5 mm³) from lung cancer cell xenografts into the right flank. For Dox-dependent tumor experiments, mice were fed with Dox-containing sugar water 7–14 days after tumor transplantations. The volume and weight of tumors were monitored and evaluated. The tumor volume was calculated using the following formula [26]: V = 0.52 × length × width².

To identify potential NSCLC-associated kinase genes, we analyzed the expression differences of the top 50 kinase genes positively regulated in lung cancer compared to the normal control in microarray data from GEO datasets (GSE19804) (Additional file 3: Fig. S1A). 7 genes (CKMT1, HKDC1, AURKA, FRK, CHEK1, EIF2AK1, and BORA) were further evaluated as candidate genes involved in NSCLC tumorigenesis excluded those genes are well studied in cancer or the value of Log₂FC < 1 (Additional file 3: Fig. S1B). To learn more about the function of these candidate genes, we performed a functional assay using small interfering RNAs (siRNAs) to knock down these genes in A549 cells respectively (Additional file 2: Table S1). The result showed that the decrease in cell proliferation is most significantly compared to the others in CKMT1-knockdown cells (Additional file 3: Fig. S1B). Additionally, we detected CKMT1 expression in NSCLC by immunohistochemistry (IHC) staining. The results showed that CKMT1 expression was markedly elevated at the protein level (68.6%; 129/188) in NSCLC tissues compared to paired normal tissues (P < 0.001; Additional file 3: Fig. S2A, B). These data suggest that CKMT1 is potentially involved in the regulation of NSCLC cell proliferation.

Next, we knocked down CKMT1 expression in PC9 or A549 cells respectively, and then overexpressed CKMT1 in H1299 cells and performed CCK8 assays to assess cell proliferation. The results showed that CKMT1 knockdown significantly suppressed the proliferation in PC9 and A549 cells, whereas ectopic overexpression of CKMT1 significantly enhanced the proliferation in H1299 cells (Fig. 1A–C). In addition, the ability of proliferation can be recovered when the CKMT1 was re-expressed in CKMT1-knockout A549 (A549 CKMT1-KO) cells through the functional rescue assay (Fig. 1D). Thus, we know that CKMT1 promotes the proliferation of lung cancer cells in vitro from the results mentioned above.

We then tested whether CKMT1 could influence the growth of NSCLC xenografts in vivo. We established the doxycycline-inducible CKMT1 overexpressed H1299 cells or knockdown A549 cells using Dox-on inducible expression system. Then we generated a xenograft model via subcutaneous transplant of tumor sections derived from subcutaneous injection of CKMT1 or shCKMT1-expression cultured cells. The results in nude mice showed that the volumes and weights of tumors generated from CKMT1-induced H1299 cells were significantly larger than those tumors without inducing CKMT1 overexpression in mice through feeding sugar water containing Dox (Fig. 1E). Consistent with the results obtained from the up-regulation experiments, the volumes and weights of tumors from the Dox group increased more slowly than those of the No Dox group that mice were fed without Dox containing sugar water (Fig. 1F). Collectively, these findings indicate that CKMT1 promotes cell proliferation in vitro and in vivo and contributes to NSCLC tumorigenesis.

To further investigate the molecular mechanisms of CKMT1 on NSCLC tumorigenesis, we performed a genome-wide transcriptome analysis using RNA-Seq in CKMT1 knockdown or knockout A549 cells (Additional file 1). The results of GO enrichment analysis indicated that the cell cycle pathway was the most significantly enriched pathway associated with the down-regulation of CKMT1 (Fig. 2A, B). Then we performed a cell cycle assay by flow cytometry to confirm the result. Compared to control cells, CKMT1 knockdown resulted in a significantly increased G0/G1 fraction and decreased S phase cell fraction, indicating G1 phase arrest (Fig. 2C). On the contrary, the proportion of cells in the G1 phase was lower than the control group after overexpressing CKMT1 in H1299 cells, while the proportion of S phase cells increased significantly (Additional file 3: Fig. S2C). Additionally, we also performed a cell cycle analysis with synchronization using serum starvation to further elucidate the role of CKMT1 in modulating the cell cycle. Compared to the control cells, we found that CKMT1 deficient cells exhibited more resistance to G1 phase arrest induced by serum starvation (73.8–82.7% vs. 61.8–75.3%). After adding serum, CKMT1 deficient cells entered the S phase from G0/G1 phase more slowly, and the proportion of these cells was significantly lower than CKMT1 wide type cells (Fig. 2D). These data suggest that CKMT1 plays an important role in regulating the G1-S phase transition in NSCLC cells.

To determine the molecular targets of CKMT1 in regulating the cell cycle, we performed a Co-IP assay for Mass Spectrometry (MS) analysis in A549 cells, and CDK4 was identified. CDK4 is a protein-serine kinase involved in the cell cycle, which interacts with cyclin D1 to form the cyclinD1-CDK4 complex, and then phosphorylates retinoblastoma (Rb) to release transcription factor E2F, thus promoting the cell from G1 phase to S phase [27]. To confirm the interactions between CKMT1 and CDK4, we performed co-IP assays and found that CKMT1 and CDK4 could interact with each other in A549 cells (Fig. 3A, B). In addition, the results of the GST pull-down assay showed that CDK4 could directly interact with purified GST-CKMT1 (Fig. 3C).

To further determine the binding fragment of CKMT1 with CDK4, we generated a series of plasmids containing Flag-tagged different truncated fragments of CKMT1 and CDK4 (Fig. 3D, E) and investigated the effects of these deletions or truncations on binding with each other in A549 cells. As known in Fig. 3D, the binding of CKMT1 with CDK4 was completely abolished by the removal of the fragment of 211–256aa, indicating that the Dbl homology domain 2 (DH2) was required for the binding of CKMT1 to CDK4, rather than the domain of creatine phosphorylation (CPS, the fragment of 257–323aa). For CDK4, we further found that the truncated N- and C-terminal variants of CDK4 could interact with CKMT1 (Fig. 3E). These results support a direct interaction between CDK4 and CKMT1, and the crucial binding areas were focused on the DH domain of CKMT1 and the N- and C-terminal of CDK4.

To investigate whether the function of CKMT1 in the cell cycle is related to its binding ability to CDK4, we performed CCK8 assays to assess cell proliferation of various deletions of CKMT1 in CKMT1-knockout A549 cells. The results showed that the ability of proliferation can be completely recovered when the C-terminal fragment deleted CKMT1 (Δ324–417aa) was re-expressed; however, less recovery was observed when the DH2 deleted CKMT1 (Δ211–256aa) was re-expressed (Fig. 3F), these data suggest the function of CKMT1 is partly dependent on the interaction with CDK4. This suggestion had been verified further by another experiment, we found that CKMT1-knockout rendered A549 cells more resistant to CDK4 inhibitor-Palbociclib as compared with CKMT1 wide type cells, which could not be affected by creatine kinase inhibitor-Cyclocreatine (Ccr) (Fig. 3G). Together, these results demonstrate that CKMT1 functions by interacting directly with CDK4, and this process was unrelated to the activity of creatine phosphorylation.

Since the recovery ability was completely abolished by the removal of mitochondrial presequence (Δ1-38aa) in CKMT1-KO cells (Fig. 3F), which means CKMT1 only exhibited the function of regulating proliferation in mitochondria, we then asked whether CDK4 may enter to mitochondria for interaction with CKMT1. For this, we carried out living cell imaging to assess. The result of confocal colocalization studies demonstrated that CDK4 (GFP) is partly located in mitochondria when added serum 2 h after synchronization in CKMT1-WT cells, however, knockout of CKMT1 abrogated localization of CDK4 to mitochondria (Fig. 4A), these means that CKMT1 plays a key role in rendering CDK4 located in mitochondria. To further confirm this, we performed mitochondrial isolation for western blot, to identify CDK4 could present in mitochondrial fractions (Fig. 4C). Of note, we also observed that the deficiency of CKMT1 markedly suppressed the nuclear translocation of CDK4 (Fig. 4B). Taken together, these results indicate that a fraction of the CDK4 proteins colocalize and interact with the CKMT1 at mitochondria, thus influencing the transition of the cell cycle.

CDK4, a cyclin-dependent kinase, is well known to phosphorylate Rb to promote G1/S cell cycle transition, and the activity of kinase is determined by its phosphorylation [28]. To better understand the regulation of CKMT1 on the activation of CDK4, we detected the phosphorylated level of CDK4 and its downstream Rb. As expected, we found the level of phosphorylated CDK4 and phosphorylated Rb decreased when we knocked down CKMT1 expression in A549 cells (Fig. 4D). In addition, Dox-induced shCKMT1-expression resulted in gradually decreased levels of phosphorylated CDK4, as well as the levels of phosphorylated Rb (Fig. 4E). These results suggest a potential role for CKMT1 in ultimately regulating the cell cycle via phosphorylation of CDK4 to activate downstream transcriptional responses.

Paclitaxel (TAX) as a tubulin inhibitor, has been used to inhibit cancer cell growth and induce G2/M cell cycle arrest [29, 30]. Thus, we asked whether tumor cells with lower CKMT1 expression might have increased sensitivity to TAX. To test this possibility, we added the G2/M cell cycle arrestor TAX to A549 control and CKMT1 knockdown cells in vitro and in vivo. Our results showed that knockdown of CKMT1 did significantly decrease the IC₅₀ of paclitaxel compared to A549 shNC cells (IC₅₀ = 137.40 ± 2.76 nM in A549 shNC vs. IC₅₀ = 18.91 ± 0.91 nM in A549 shCKMT1-1 and 11.61 ± 1.34 nM in A549 shCKMT1-2, P < 0.01). Meanwhile, CKMT1 overexpressed cells (IC₅₀ = 369.81 ± 10.37 nM) were 4.5-fold more resistive to paclitaxel than control vector cells (IC₅₀ = 82.90 ± 3.94 nM) (Fig. 5A). In animal experiments, the tumor volume and tumor weight in the combination of CKMT1-KO and TAX treatment group were significantly reduced after TAX treatment compared to the control group (Figs. 5B). Therefore, CKMT1 affects the anti-tumor effect of paclitaxel and may be a potential target to improve the sensitivity to chemotherapy.