2.1 Patients and clinical tissue specimens

The inclusion criteria were as follows: (1) patients who were diagnosed as LUAD on pathology and underwent epidermal growth factor receptor (EGFR) gene testing; (2) availability of enough tumor tissue and paired normal lung tissue acquired by surgery for further experiments in additional to routine clinical requirements; (3) patients who received comprehensive treatment according to clinical needs; (4) patients who agreed to participate in this study and were followed up regularly. LUAD patients who were unwilling to participate in the study, lost to follow up, or rejected systemic therapy were excluded from this research. In total, 71 LUAD patients were included. The clinical data of these patients were collected, and the LUAD tissues and paired adjacent normal lung tissues were used for immunohistochemical analysis. This study was approved by the Institution Review Board of Huashan Hospital, Fudan University, China (Permit No. KY2016-396), and each patient signed the informed consent form.

2.2 Cell lines and reagents

The human lung cancer cell lines NCI-H1975, HCC827, PC-9, and immortalized bronchial epithelial cell line BEAS-2B were obtained from the Cell Bank of Shanghai Institute for Biological Sciences, Chinese Academy of Sciences. NCI-H1299, NCI-H1650, and A549 were purchased from ATCC. No cross-contamination was observed in all of those cell lines, which was detected by polymerase chain reaction (PCR) of short tandem repeat DNA fingerprints. Cells were cultured in a complete growth medium with DMEM or RPMI 1640 medium containing 10% fetal bovine serum (FBS) and 100 U/mL penicillin–streptomycin. The cell dishes were placed in a humidified incubator with 5% CO₂ maintained at 37°C, and passed on at 80%–90% confluence. Cells were split every 3–4 days or as needed. PCR was performed to screen for Mycoplasma contamination of cultured cells.

2.3 Immunohistochemistry

The procedures of immunohistochemistry were as described in a previous study.²² Generally, the tissues were fixed with 10% formalin, embedded in paraffin, and cut into a series of sections. Then, the sections were dewaxed, incubated with 3% H₂O₂ for 10 min in the dark, and blocked in 10% goat serum for 2 h. After incubation with primary CDK10 antibody (1:100, Cell Signaling Technology, #36106), ETS2 (1:100, Abcam, ab219948), anti-phospho-Ser217/221-MEK1/2 (1:100, Cell Signaling Technology, #9154), or phospho-Thr202/Tyr204-ERK1/2 (1:100, Cell Signaling Technology, #4370) at 4°C overnight, the sections were incubated with biotinylated secondary antibody (Cell Signaling Technology), stained with 3,3-diaminobenzidine and counterstained with hematoxylin. All slides were observed and photographed by an Olympus microscope (Nikon) under the same exposure conditions. The expression levels of CDK10, ETS2, p-MEK1/2, and p-ERK1/2 were scored by an experienced pathologist by an immunoreactive score, as described before.²³

2.4 Gene knockdown (KD) and overexpression (OE) using lentivirus vectors

The short hairpin RNA (shRNA) oligonucleotides targeting CDK10 were designed and synthesized by Synbio Technologies, China. The sequences of CDK10 shRNA are shown in Supporting Information Table S3. Next, the paired shRNA oligos and control scrambled sequences were cloned into pGreen-Puro-CMV plasmid (Public Protein/Plasmid Library). The full-length CDK10 isoform a (NM_052988.5) was synthesized (GenScript) and cloned into the pCDH-CMV-MCS-EF1-Puro plasmid (Public Protein/Plasmid Library). The complementary DNA (cDNA) of ETS2 (NM_001256295.1) was cloned into the vector of EX-NEG-Lv183 (Gene Copoeia Company). These plasmids were cotransfected with psPAX2 and pMD2.G plasmids into HEK293T cells. The recombinant lentiviruses in the cell culture medium were collected and used to transfect the H1299, PC-9, and A549 cells. The KD and OE efficiencies of CDK10 in the transfected cells were verified by Western blot.

2.5 Quantitative reverse transcription-PCR (qRT-PCR) analysis

Total RNA was extracted using TRIzol reagent (Invitrogen) in accordance with the manufacturer's protocol. Then, 1 μg of RNA was reverse-transcribed into cDNA using a PrimeScript Reverse Transcriptase reagent kit (Takara). The qRT-PCR was conducted in C1000 Thermo Cycler (Bio-Rad), using the mixture of cDNA, SYBR Premix Ex Taq (Takara), and primers (Sangon Biotech). The primer sequences of the relative genes are shown in Supporting Information Table S2. The relative messenger RNA (mRNA) expression level was calculated as follows: ΔCt = Ct_gene − Ct_{Internal reference}; ΔΔCt = ΔCt_{experimental group} − ΔCt_{control group}; the relative mRNA expression level = 2^−ΔΔCt.

2.6 Immunoblotting

The protein sample was extracted from the cells by RIPA lysis buffer (Beyond, P0013B), and then, its concentration was examined using BCA kits (Thermo Fisher Scientific, #23225). After denaturation in a metal bath at 100°C for 15 min, the proteins were separated in 10% sodium dodecyl-sulfate polyacrylamide gel electrophoresis gels using electrophoresis tanks (Bio-Rad) and transferred to polyvinylidene difluoride transfer membranes. Five percent bovine serum albumin in Tris–buffered saline (TBST) was applied to block the blots for 1 h at room temperature. Primary antibodies were added and incubated overnight at 4°C. The following primary antibodies were used in this experiment: anti-CDK10 (Cell Signaling Technology, #36106), anti-E-cadherin (Cell Signaling Technology, #3195), anti-N-Cadherin (Cell Signaling Technology, #4061), anti-ETS2 (Abcam, ab219948), anti-phospho-Thr202/Tyr204-ERK1/2 (Cell Signaling Technology, #8544), anti-ERK1/2 (Cell Signaling Technology, #9101), anti-phospho-Ser217/221-MEK1/2 (Cell Signaling Technology, #9154), anti-MEK1/2 (Cell Signaling Technology, #4694), anti-MMP2 (Abcam, ab92536), anti-MMP9 (Abcam, ab76003), anti-C-Raf (Erpantech, AB-06-3811), anti-GAPDH (Cell Signaling Technology, #97166), anti-β-Tubulin (Cell Signaling Technology, #2146), and anti-Vimentin (Cell Signaling Technology, #5741). After washing out of the primary antibody by TBST solution, the membranes were incubated with horseradish peroxidase-conjugated secondary antibody for 1 h, followed by a reaction with ECL chemiluminescence reagent (Amersham Life Science). Finally, the blots were visualized and scanned in a chemiluminescence imaging system (Tanon).

2.7 Transwell migration and invasion assay

In the cell migration assay, cells in the culture medium without FBS were seeded on the upper chamber (Corning, 8 μm), and 1 mL of complete cell culture medium containing 10% FBS was added into the lower chamber in a 24-well plate as well. After incubation for 14–16 h, the cells attached to the upper chamber were fixed with 4% paraformaldehyde and stained with 0.1% crystal violet. The cells on the upper side of the membrane were gently removed by cotton swab, while the cells on the lower side of the membrane were photographed and counted in five random fields. In the cell invasion assay, each chamber was pretreated with 40 μg matrigel (Corning) in line with the manufacturer's protocol, and the remaining processes were similar to those in the transwell migration assay. All experiments were repeated for at least three times.

2.8 Immunoprecipitation

pCDH-CDK10-Myc (6 μg) and an empty vector (6 μg) were transfected into H1299 cells separately. Twelve hours later, MG132 (10 μM; MedChemExpress) was added to inhibit the degradation of proteins. After incubation for another 24 h, the protein samples were prepared in 0.5% NP-40 lysis buffer containing protease and phosphatase inhibitors and were further lyzed by an ultrasonic cell crusher (Scientz). The lysate (1 mg/sample) was incubated with 25 μL of anti-Myc-tag magnetic beads for 3 h at 4°C on a rotary shaker. Then, the beads were washed with TBST, boiled for 10 min, and subjected to Western blot assay.

2.9 Immunofluorescence imaging

H1299 cells in 50%–70% confluence were fixed with 4% paraformaldehyde, blocked with 5% BSA, incubated with primary antibodies against CDK10 (1:100, Thermo Fisher Scientific, PA5-96192) and ETS2 (1:50, SantaCruz, sc-365666) overnight at 4°C, and then incubated with Alexa Fluor 488-conjugated and Dylight 550-conjugated secondary antibodies for 1 h. Then, 4ʹ,6-diamidino-2-phenylindole dye was applied to stain the nucleus, and cells were observed and photographed under a fluorescence microscope.

2.10 RNA interference

Small interference RNAs (siRNAs) targeting ETS2 and control scrambled siRNAs were designed and synthesized by Genepharma. Cells were transfected with ETS2 siRNAs and control siRNAs using Lipofectamine 3000 (Invitrogen). The sequences of siRNAs targeting ETS2 and negative control are shown in Supporting Information Table S3. The qRT-PCR and Western blot were conducted to verify the KD efficiency of ETS2.

2.11 In vivo model of LUAD metastasis

The lentiviral vector expressing firefly-luciferase (Genechem) was transfected into CDK10-KD and control H1299 cells. Twelve female BALB/c nude mice aged 4 weeks were randomly divided into two groups, injected with shCDK10 H1299 cells and control cells (5 × 10⁵ cells in 200 μL PBS/mouse) via the caudal vein, respectively. Six weeks later, metastatic lesions in distant lungs were detected by bioluminescence imaging after intraperitoneal injection with d-luciferin (150 mg d-luciferin/kg body weight). The mice were killed, and the lung was isolated, fixed with 10% formalin, embedded in paraffin, and cut into a series of continuous sections at an interval of 50 μm. Each slide was then processed with hematoxylin–eosin (HE) staining and the metastatic lesions were counted. All of the mice were raised in a pathogen-free animal laboratory animal housing with free access to food and water, in accordance with the National Institutes of Health guide. All animal experiments were approved by the Fudan University Veterinary Medicine Animal Care and Use Committee.

2.12 Statistical analysis

ImageJ software was used for grayscale scanning of Western blots, and the GraphPad Prism 8.0 software was used for statistical analysis. The data of continuous variables were expressed as mean ± standard deviation (SD). The t test was used to compare the data between two groups, and the analysis of variance was used to compare the data between multiple groups if these data were homogeneous and distributed normally. If not, the continuous variables were analyzed by rank-sum test. Log-rank test was used to analyze the differences between Kaplan–Meier survival curves. If p was <0.05, the difference was considered as significant.

3.1 CDK10 downregulation correlates with metastasis and poor overall survival in LUAD patients

In this study, the paired tumor tissues, adjacent normal lung tissue specimens, and metastatic lesions in lymph nodes from five T1aN1M0 and five T1aN0M0 stage LUAD patients were collected for the next-generation sequencing (NGS). Analysis of the NGS data resulted in the identification of CDK10, which harbored a missense mutation in tumors from 3 of 5 enrolled patients but not in the paired normal tissues (Supporting Information Table S1). The correlation of CDK10 expression with the clinical characteristics of 71 patients with LUAD was next analyzed. Immunochemical staining indicated that the levels of CDK10 protein in tumor tissues were downregulated compared with those in the paired normal lung tissues (Figure 1A,B). As shown in Table 1, CDK10 was lower in patients with stages N2, M1, and TNM stage II/IV than in patients with stages N0, M0, and TNM stage I, respectively. Besides, CDK10 was also lower in LUAD of pathologic grade III versus grades I and II. Consistently, data on UALCAN.²⁴ showed a reduced mRNA level of CDK10 in LUAD patients with N1 and N2 stages compared with those with N0 stage (Figure 1C). Patients with TNM stages II and III also had lower mRNA levels of CDK10 than those with TNM stage I (Figure 1D). Kaplan–Meier survival curve analysis indicated a shorter overall survival in patients with lower protein expression of CDK10 (hazard ratio [HR] = 0.5167; 95% confidence interval [CI], 0.2835–0.9415; Log-rank P = 0.0405; Figure 1E). We further examined the association of CDK10 mRNA levels with the prognosis of LUAD patients in the public Kaplan–Meier Plotter database,²⁵ and verified shorter overall survival in patients with lower CDK10 mRNA expression (HR = 0.40; 95% CI, 0.29–0.55; Log-rank P = 8.2e⁻⁹; Figure 1F). Thus, CDK10 was downregulated in LUAD, and the downregulation of CDK10 level was significantly correlated with increased metastatic capacity of tumors and poor prognosis of LUAD patients.

3.2 CDK10 inhibits the migration and invasion abilities of LUAD cells

The expression of CDK10 was examined in lung cancer cell lines and the normal lung epithelial cell line, BEAS-2B. As shown in Figure 2A,B, the mRNA and protein levels of CDK10 were significantly higher than those in the LUAD cell lines H1299 and H1975 than in the normal lung epithelial cell line, BEAS-2B. The KD and OE of CDK10 in LUAD cells were achieved by lentivirus-delivered shRNAs or CDK10 cDNA (Figure 2C–E). CCK8 and EdU assays indicated no significant effect of CDK10 on the viability of LUAD cells (Supporting Information Figure S1A,B), which was consistent with the finding that CDK10 KD failed to induce cell cycle arrest in LUAD cells (Supporting Information Figure S1C). Transwell migration and invasion assays showed that CDK10 KD enhanced the migration and invasion capabilities of H1299 and PC-9 cells (Figure 2F,G), and CDK10 OE impaired the migration and invasion capabilities of A549 cells (Figure 2H). Consistently, wound healing assay indicated that CDK10 KD promoted the migration of H1299 cells, and CDK10 OE inhibited the migration of A549 cells (Supporting Information Figure S2). These data suggest that CDK10 plays a suppressive role in the motility and invasiveness of LUAD cells.

3.3 CDK10 represses the formation of in vivo metastatic foci by LUAD cells

A xenograft metastatic model of LUAD was established in BALB/c nude mice by injection of H1299 cells stably expressing firefly-luciferase via the caudal vein. The results of bioluminescence imaging showed that CDK10 KD significantly enhanced the formation of metastatic lesions (Figure 3A). Consistently, HE staining detected more metastatic lesions in the lungs in mice injected with CDK10-KD H1299 cells than those in control cells (Figure 3B). Therefore, CDK10 counteracts the development of metastases of LUAD cells in vivo.

3.4 CDK10 interacts with ETS2 and inhibits the degradation of ETS2 protein

A previous study has reported the interaction between CDK10 and ETS2 proteins in STAR syndrome.¹⁵ ETS2 is considered as an oncoprotein, which has been reported to enhance the tumorigenesis and progression of various types of malignancies.²⁶ ETS2 could act as a transcription factor through binding with the promoters of target genes including C-RAF and matrix metalloproteinases (MMPs), both of which are involved in the multistep processes of metastasis.^{15, 27} Thus, we speculated that the interaction of CDK10 with ETS2 might be involved in the regulation of LUAD metastasis. Immunoprecipitation verified the interaction of CDK10 protein with ETS2 protein in H1299 cells (Figure 4A). Then, dual immunofluorescence staining showed the colocalization of CDK10 and ETS2 proteins in H1299 cells, especially abundant in the nucleus (Figure 4B). No obvious effect of CDK10 KD on the synthesis of ETS2 protein has been found (Figure 4C). However, the silencing of CDK10 in H1299 cells inhibited the degradation and thereby prolonged the half-life time of ETS2 protein (Figure 4D), with the half-life time of 1.178 h and 1.836 h in control and CDK10-KD H1299 cells when treated with CHX (100 μg/mL), respectively. Accordingly, CDK10 might suppress the metastasis of LUAD cells by promoting the degradation of ETS2 protein.

3.5 CDK10 inhibits ETS2/c-Raf/p-MEK/p-ERK signaling, MMP2/9 and impedes EMT of LUAD

The target genes of ETS2, including C-RAF, MMP2, and MMP9, were increased in CDK10-KD H1299 cells (Figure 5A). The c-Raf/p-MEK/p-ERK pathway is widely activated in malignant tumors and could enhance the metastasis of malignant tumors through EMT.²⁸ We found that CDK10 KD increased the protein level of ETS2, and induced the increased expression of MMP2/9 and activation of the c-Raf/p-MEK/p-ERK/EMT signaling cascade in H1299 cells (Figure 5B). Consistently, the immunohistochemical staining in the in vivo model showed increased protein levels of ETS2, p-MEK, and p-ERK in metastatic LUAD lesions in the CDK10-KD group (Figure 3C).

Conversely, the downregulation of ETS2 and inhibitions of MMP2/9 and the c-Raf/p-MEK/p-ERK/EMT pathway were found in CDK10-overexpressed A549 cells compared with control cells (Figure 5C). To test whether ETS2 was a functional target of CDK10, we further knocked down ETS2 in H1299 cells (Figure 6A,B). Transwell migration assay indicated that the effect of CDK10 silencing on cell motility could be blocked by ETS2 KD (Figure 6C). Conversely, while CDK10 OE in A549 cells impaired the capability of invasion, this was rescued by further OE of ETS2 (Figure 6D,E). Moreover, ETS2 deficiency prevented the activation of c-Raf/p-MEK/p-ERK signaling and impaired EMT induced by CDK10 KD, as evidenced by the downregulation of mesenchymal marker N-cadherin and vimentin (Figure 6F). These results showed that ETS2 was a key downstream effector for the inhibitory role of CDK10 on LUAD metastasis. Taken together, these data suggested CDK10 suppressed metastasis of LUAD through ablating ETS2 transactivation of c-Raf and subsequently repressing the p-MEK/p-ERK pathway and EMT of malignant cells.

3.6 ETS2 and the c-Raf/p-MEK/p-ERK pathway constitute a positive feedback loop in CDK10-deficient LUAD cells

C-Raf/p-MEK/p-ERK signaling represents a canonical intracellular pathway responsible for the regulation of several genes. In line with the aforementioned findings that it plays a key role downstream of CDK10 to regulate the malignant phenotypes of lung cancer cells, we observed that PD98095, a specific inhibitor of MEK,²⁹ reversed the effect of CDK10 KD on the migration of cells (Figure 7A), and inhibited EMT induced by CDK10 ablation (Figure 7B). In addition, decreased expression of ETS2 and MMP2/9 was found in H1299 cells after treatment with PD98095 (40 μM). Consistently, the mRNA levels of MMP2 and MMP9 were also downregulated, upon treatment of cells with PD98095^{27, 30} (Figure 7C). These findings revealed a positive feedback regulation of ETS2 and MMP2/9 by the p-MEK/p-ERK pathway, which probably contributed to the metastasis of CDK10-deficient LUAD.