Distinct Regulation of ASCL1 by the Cell Cycle and Chemotherapy in Small Cell Lung Cancer

Abstract

Small cell lung cancer (SCLC) is an aggressive and lethal malignancy. Achaete-scute homolog 1 (ASCL1) is essential for the initiation of SCLC in mice and the development of pulmonary neuroendocrine cells (PNEC), which are the major cells of origin for SCLC. However, the regulatory mechanism of ASCL1 in SCLC remains elusive. Here, we found that ASCL1 expression gradually increases as the tumors grow in a mouse SCLC model, and is regulated by the cell cycle. Mechanistically, CDK2–CyclinA2 complex phosphorylates ASCL1, which results in increased proteasome-mediated ASCL1 protein degradation by E3 ubiquitin ligase HUWE1 during mitosis. TCF3 promotes the multisite phosphorylation of ASCL1 through the CDK2–CyclinA2 complex and the interaction between ASCL1 and TCF3 protects ASCL1 from degradation. The dissociation of TCF3 from ASCL1 during mitosis accelerates the degradation of ASCL1. In addition, chemotherapy drugs greatly reduce the transcription of ASCL1 in SCLC cells. Depletion of ASCL1 sensitizes SCLC cells to chemotherapy drugs. Together, our study demonstrates that ASCL1 is a cell-cycle–regulated protein and provides a theoretical basis for applying cell-cycle–related antitumor drugs in SCLC treatment.

Implications:Our study revealed a novel regulatory mechanism of ASCL1 by cell cycle and chemotherapy drugs in SCLC. Treating patients with SCLC with a combination of ASCL1-targeting therapy and chemotherapy drugs could potentially be beneficial.

Introduction

Small cell lung cancer (SCLC) is a highly lethal malignancy, accounting for about 10% to 15% of lung cancers, and its 5-year survival rate is less than 7% (1). Compared with NSCLC, SCLC develops rapidly and metastasizes at early stage (2). About 60% to 70% of patients are already in the extensive stage at the initial diagnosis (3). At present, etoposide and cisplatin (EC) are the first-line standard treatment drugs for SCLC (4). However, drug resistance is inevitable in the treatment of SCLC. In the past 40 years, there has been no significant progress in SCLC treatment. Immune-checkpoint inhibitors are currently approved for first- and third-line treatment for the patients with extensive-stage and relapsed SCLC (5). Traditional chemotherapy combined with immunotherapy improves the overall survival, but it is only beneficial for the minority of patients with SCLC (6).

Pulmonary neuroendocrine cells (PNEC) are considered as the primary cells of origin of SCLC (7, 8). In addition, few SPC-positive alveolar type II cells have the ability to form SCLC (8). SCLC tumor genome is unstable with inactivating mutations in the TP53 and RB1 genes. Loss of RB1 and TP53, and amplification of the oncogene MYC lead to an abnormally rapid proliferation of SCLC cells, resulting in replication stress. Therefore, SCLC cells are dependent on a complete DNA damage response (DDR) system. Any DDR or cell-cycle checkpoint instability leads to cell death (9, 10). Genomics and proteomics profiling results show that DDR has an abnormal response in SCLC with high expression of PARP, ATR, CHK1/2, and ATM (11, 12). Therefore, targeting DDR has great potential in SCLC therapy. Inhibition of CDK7 in SCLC affects the cell cycle and leads to micronuclei formation, increasing genomic instability. Combining CDK7 inhibitors with anti-PD1 antibodies and chemotherapy drugs enhances the effect of tumor treatment (13).

On the basis of the human SCLC tumor samples, cancer cell lines, and reported engineered mouse models, SCLC can be classified into four main subtypes by the expression of specific transcription factors, including achaete-scute homolog 1 (ASCL1), neurogenic differentiation factor 1 (NeuroD1), POU class 2 homeobox 3 (POU2F3), and yes-associated protein 1 (YAP1; ref. 14). Most SCLC cell lines and tumors (∼75%) express ASCL1. Approximately 15% of SCLC are NeuroD1 positive. ASCL1 and NeuroD1 subtypes have neuroendocrine features, and the other two subtypes have non-neuroendocrine features (15). ASCL1 belongs to the achaete-scute family, and has a basic-helix-loop-helix (bHLH) domain for DNA-binding and dimerization shared by other proneural genes (16, 17). When ASCL1 exerts transcription factor activity, it has to form heterodimers with other bHLH proteins, such as E12 and E47, encoded by E2A gene (18). Notch signaling inhibits ASCL1 expression through HES1 and HES5. HES1 and HES5 directly bind to the promoter of ASCL1 to repress its transcription (19). In addition, HES1 and HES5 interact with E47 to impair the interaction between ASCL1 and E47. Disruption of the ASCL1 gene fails to form PNECs in mice (20). In addition, ASCL1 is essential for mouse SCLC tumorigenesis (15) and human SCLC cell survival (21). Although ASCL1 is critical for SCLC tumorigenesis, the regulatory mechanism of ASCL1 in SCLC is currently poorly investigated.

Here, we found that ASCL1 protein level in SCLC cells is regulated by the cell cycle. Phosphorylation of ASCL1 by CDK2–CyclinA2 complex promotes ASCL1 degradation in mitosis by E3 ligase HUWE1. Moreover, reduced ASCL1 expression makes SCLC cells more sensitive to chemotherapy drugs.

Materials and Methods

Cell culture, transfection, and lentiviral transduction

The HEK293T and HeLa cells were cultured in DMEM (GIBCO) medium with 10% FBS (GIBCO) and 1% penicillin/streptomycin (Beyotime Biotechnology). NCI-H69, NCI-H209, and primary mouse SCLC cells were cultured in RPMI1640 medium with 10% FBS (GIBCO) and 1% penicillin/streptomycin (Beyotime Biotechnology). All cells were grown at 37°C with 5% CO₂. HEK293T, HeLa, NCI-H69, DMS53, and NCI-H209 were obtained from ATCC. These cell lines have been authenticated by using short tandem repeat profiling analysis and were mycoplasma-free. For replicate experiments, cells were cultured for 1 month at most. Polyethyleneimine (PEI; Polyscience) and Lippo 6000 (Beyotime Biotechnology) transfection reagents were used for plasmid transfection. For lentiviral production, the mixture of pMD2.G, pSPAX2, and lentiviral plasmids (1:3:4) was introduced into HEK293T. Media containing lentivirus were collected at 48 and 72 hours after transfection. SCLC cells were resuspended in 1 mL lentivirus with 8 μg/mL polybrene in a 6-well plate and centrifuged at 900 × g for 2 hours. The lentivirus was removed 24 hours later.

Animal studies

Nude mice (male, 6 weeks old) were purchased from Shanghai SLAC Laboratory Animal Company. Xenograft models were established by subcutaneously injecting 1 × 10⁷ NCI-H209 cells in Matrigel (Corning) into bilateral flanks of nude mice. For knockdown ASCL1 tumor formation assay, when tumors were visible, the mice were monitored every 5 days. For paclitaxel (PTX) in combination with EC treatment, when the tumor reached a mean diameter of around 0.5 cm, mice were assigned to four groups (n = 5/group): (i) vehicle; (ii) PTX (10 mg/kg, i.p.); (iii) EC (cisplatin 7.5 mg/kg, etoposide 15 mg/kg, i.p.); and (iv) combination of PTX and EC. Treatment with PTX was repeated every 3 days. Acute EC treatment was carried out in the last week before harvest the tumor. In the last week of acute EC treatment, mice were treated with cisplatin on day 1, and a combination of EC on days 2 and 4 (22). Tumor volume was calculated: tumor volume (mm³) = length × (width)²/2. The animal studies were approved by Zhejiang University Animal Care and Use Committee.

Plasmids and reagents

Expression plasmids for ASCL1, CDK2, CyclinA2, and TCF3 were generated by standard molecular biology techniques using cDNAs from NCI-H69 and HEK293T cells as templates. cDNAs were constructed into pCDNA3 or pLVX lentiviral vector. sgRNA, shRNA, and siRNA target sequences are listed in the Supplementary Table S1. MG132 and small-molecule inhibitors were obtained from Selleck Chemicals. p-nitrobenzyl mesylate (PNBM) and ATP-γ-S were obtained from Abcam. Cisplatin and etoposide were purchased from TargetMol.

RNA extraction and qRT-PCR

Total RNA from cultured cells was extracted using the RNAiso Plus reagent (Takara). Use PrimeScript RT Master Mix (Takara) to generate cDNA. Quantitative real-time PCR was performed using SYBR Green PCR Master Mix (YEASEN and Vazyme). Relative gene expression levels were normalized by RPLP0 or GAPDH. PCR primers were listed in Supplementary Table S2.

Immunoprecipitation, immunoblot, immunofluorescence

Cells were washed with PBS and lysed with lysis buffer (150 mmol/L NaCl, 50 mmol/L Tris-HCL pH 7.5, 0.5% NP-40, 10% glycerol, 1 mmol/L PMSF, 1 mmol/L DTT). Cell lysates were incubated with anti-FLAG, anti-MYC, or anti-ASCL1 conjugated magnetic beads overnight at 4°C. Beads were washed with lysis buffer for three times and immunoprecipitants were eluted with 1× SDS loading buffer. The proteins were separated by SDS-PAGE and then transferred to PVDF membrane (EMD Milipore). The images of immunoblot were captured by ChemiScope5600 (Clinx) with ECL substrate. ImageJ was used for quantitative analysis. For immunofluorescence, HeLa cells were seeded on the poly-lysine treated glass coverslips in 24-well plates. After transfection, cells were washed with PBS and then fixed with 4% PFA for 20 minutes at room temperature, followed by permeabilization with 0.5% Triton X-100 for 5 minutes. The cells were blocked with 3% BSA in PBST (PBS, 0.1% Tween 20) for 40 minutes and incubated with primary antibodies for 2 hours at room temperature and followed by incubation with Alexa Fluor-labeled secondary antibodies for 1 hour at room temperature. Nuclei were stained with DAPI at room temperature for 10 minutes. The cells were mounted with 50% glycerol (PBS) and the images were captured with a laser scanning confocal microscope (Zeiss). Antibody information was listed in Supplementary Table S3.

IHC

Mouse lungs were perfused with PBS through the right ventricle to remove the blood and inflated with 4 mL of 4% PFA and fixed for 6 hours at 4°C in 4% PFA. After the process of dehydration, tissues were embedded in paraffin. The tissues were sectioned at 6 μmol/L thickness for IHC staining. The staining procedure was described as previously (23).

Dual-luciferase reporter assay

For luciferase assay, cells were seeded in a 24-well plate and transfected with pGL4.20-basic, Dll1, or DLL3 luciferase reporter plasmids together with pRL-TK vector as the internal control. Cells were lysed 48 or 72 hours after transfection, and luciferase activity was measured according to the manufacturer's instructions of the Dual Luciferase Assay Kit (Promega).

In vitro kinase assay

MYC-tagged CyclinA2, FLAG-tagged CDK2, and mCherry-tagged ASCL1 were immunoprecipitated from transfected HEK293T cells using corresponding antibody and protein A/G magnetic beads. Beads were washed with lysis buffer (150 mmol/L NaCl, 50 mmol/L Tris-HCL pH 7.5, 0.5% NP-40, 10% glycerol,1 mmol/L PMSF, 1 mmol/L DTT) three times at 4°C, then washed with wash buffer (200 mmol/L NaCl and 40 mmol/L HEPES) once and washed with kinase assay buffer (50 mmol/L KAC, 5 mmol/L MgCl₂, and 30 mmol/L HEPES) once. Kinase and substrate were mixed in the presence of ATP-γ-S (500 μmol/L; Abcam). The mixtures were shaken at 1,200 rpm at 30°C for 1 hour. One μL of 0.5 M EDTA (final concentration 20 mmol/L) was added to the system and the mixture continued to shake for 5 minutes to terminate the reaction. PNBM (Abcam) was added into the mixture to a final concentration at 2.5 mmol/L and continued to vibrate at 25°C for 40 minutes. The phosphorylation results were shown by immunoblot with anti-thiophosphate ester antibody (Abcam).

Statistical analysis

Multiple groups comparison was performed by one-way ANOVA followed by Tukey test or one-way ANOVA with Dunnett test. The quantitative results were presented as mean ± SD from a representative of at least three independent experiments. P values < 0.05 were considered to be significant. ^∗, P < 0.05; ^∗∗, P < 0.01; ^∗∗∗, P < 0.001. Data analyses were performed with GraphPad Prism 6. No statistical methods were used to predetermine sample size.

Data availability

The data generated in this study are available upon request from the corresponding authors.

Results

ASCL1 protein is degraded through the ubiquitin–proteasome pathway during mitosis

ASCL1 is required to establish the lineage of PNECs and is necessary for the survival of SCLC (15). We utilized the Cgrp^CreER/+; p53^f/f; Rb^f/f; Pten^f/f; Rosa26^mTmG/+ mouse SCLC model to examine the dynamic expression of ASCL1 during the development of SCLC. When the mice grew to 6 weeks of age, tamoxifen was injected intraperitoneally to induce tumor formation. In this model, all tumor cells initiated from PNECs expressed GFP. We collected the mouse lungs on the 7th, 14th, 30th, and 51st days after the tamoxifen injection. Immunofluorescence analysis showed that ASCL1 protein expression gradually increased during the progression of tumors (Fig. 1A). These result suggest that ASCL1 may play a role in the development of SCLC. The increased ASCL1 protein levels correlated with rapid cell proliferation, which led us to speculate whether ASCL1 is intrinsically regulated during cell-cycle progression. We treated human SCLC cells NCI-H69 with chemical inhibitors and found that three cell-cycle related inhibitors, Nocodazole, Taxol, and PLK1 inhibitor BI2536, significantly inhibited ASCL1 expression (Fig. 1B and C). TSA has been reported to reduced ASCL1 transcription through inducing Notch activity (24). The three different cell-cycle inhibitors effectively arrested cells in mitosis. Reduced ASCL1 protein levels in G₂–M phase was further confirmed in ASCL1-expressing human SCLC cells NCI-H209 and mouse SCLC cells derived from Cgrp^CreER/+; p53^f/f; Rb^f/f; Pten^f/f; Rosa26^mTmG/+ SCLC mouse model (Fig. 1D). Next, we synchronized NCI-H69 cells at the G₂–M phase with nocodazole, followed by release, and collected cells every 4 hours. The immunoblot results showed that the expression of ASCL1 protein recovered when the cells gradually exited the M phase (Fig. 1E). In addition, the same phenomenon was observed when the arrested cells released from the late G2 by RO3306 treatment (Supplementary Figs. S1A–S1C). These results indicate that ASCL1 protein level is regulated by the cell cycle.

Figure 1.

ASCL1 protein is degraded through the ubiquitin–proteasome pathway during mitosis. A, ASCL1 immunofluorescence staining on SCLC tumors from Cgrp^CreER/+; p53^f/f; Rb^f/f; Pten^f/f; Rosa26^mTmG/+ adult mice at indicated times after tamoxifen injection. Scale bar, 20 μm. Bottom: Quantification of immunofluorescence intensity of ASCL1 (n = 20 cells/time point). Data are presented as means ± SD. **, P < 0.01; ***, P < 0.001. B, Immunoblot analysis of NCI-H69 cells treated with different inhibitors. Nocodazole 100 ng/mL, Taxol 1 μmol/L, BI2536 20 nmol/L, TSA 1 μmol/L. C, Immunoblot analysis of NCI-H69 cells treated with Taxol at indicated times. The increased PLK1 expression indicated cells were in mitosis. D, Immunoblot analysis of NCI-H69, NCI-H209, and mouse SCLC (mSCLC) cell lines treated with nocodazole (100 ng/mL). E, NCI-H69 cells were treated with nocodazole for 26 hours to be blocked in mitosis, and then released at indicated times followed by immunoblot analysis of ASCL1 proteins. F, Immunoblot analysis of ASCL1 expression in NCI-H209 cells with pretreatment of nocodazole for 24 hours followed by treatment with MG132 or Chlq (chloroquine). G, HEK293T cells were transfected with indicated plasmids for 24 hours and treated with nocodazole for another 24 hours, then harvested for ubiquitination analysis. MG132, 20 μmol/L.

Both gene transcription and protein posttranslational modifications regulate protein expression. We first synchronized NCI-H209 cells at G₂–M phase by nocodazole treatment and examined the mRNA level of ASCL1. ASCL1 mRNA levels remained unchanged during the cell cycle in NCI-H209 cells, whereas the increased CyclinB1 expression indicated that the cells were in mitosis (Supplementary Fig. S1D). Intracellular protein degradation is mainly through the ubiquitin–proteasome pathway or the lysosome pathway. Next, we treated synchronized cells with the proteasome inhibitor MG132 or lysosome inhibitor chloroquine. The results showed that MG132 completely prevented the degradation of ASCL1 after synchronization, whereas chloroquine did not (Fig. 1F). We examined the ubiquitin modification of ASCL1 in mitosis. Consistently, nocodazole treatment greatly enhanced the ubiquitination of ASCL1 (Fig. 1G). Subsequently, we generated a stable cell line expressing ASCL1 in HEK293T cells and then treated the cells with nocodazole in a time course fashion. We found that ASCL1 protein level was gradually reduced when cells were blocked at G₂–M phase (Supplementary Fig. S1E). Together, these results suggest that ASCL1 protein is degraded through the ubiquitin–proteasome pathway during mitosis.

Phosphorylation of ASCL1 reduces its protein stability by CDK2–CyclinA2, but promotes its transcriptional activity

We found that ASCL1 proteins in SCLC cells usually display two bands on SDS-PAGE by the Western blot analysis. The treatment of SCLC cell lysates with λ-phosphatase (λ-PPase) resulted in the diminishment of the upper band (Supplementary Fig. S2A). Previous study showed that ASCL1 was phosphorylated by CDK2–CyclinA2 and CDK2–CyclinE complex (25). We were wondering whether the cell-cycle-related protein kinases may regulate the protein stability of ASCL1 by phosphorylating ASCL1. To test this hypothesis, we synchronized NCI-H209 cells in mitosis with nocodazole and then treated the synchronized cells with CDK kinase inhibitors for 50 minutes before harvesting the cells. We found that CDK1/2 inhibitor BMS265246 partially rescued the degradation of ASCL1 during mitosis (Fig. 2A and B). In addition, mouse SCLC cells displayed a similar response to BMS265246 (Fig. 2C). We overexpressed tagged ASCL1, CyclinA2, and CDK2 in HEK293T cells. The coimmunoprecipitation assay showed that ASCL1 interacted with CyclinA2 and CDK2 (Fig. 2D and E).

Figure 2.

Phosphorylation of ASCL1 by CDK2–CyclinA2 reduces its protein stability but promotes its transcriptional activity. A, Nocodazole-arrested mitotic NCI-H209 cells were treated with indicated inhibitors for 50 minutes and then were harvested for immunoblot analysis. B and C, Immunoblot analysis of ASCL1 in nocodazole-arrested mitotic NCI-H69 (B) cells and mSCLC cells (C) treated with BMS265246 for 2 hours before harvest. D and E, Co-immunoprecipitation assay in HEK293T cells transfected with indicated plasmids. F and G, HEK293T cells were transfected with indicated plasmids, and cells were harvested for immunoblot analysis (F), or cell lysates were subjected to lambda phosphatase treatment and immunoblotting (G). H, Workflow of the in vitro kinase assay (left). Immunoblot analysis of the phosphorylation of ASCL1 by CDK2 and CyclinA2 (right). I, Immunoblot analysis of wild-type ASCL1 and ASCL1–5SA mutant with or without CDK2–CyclinA2 complex in HEK293T cells. J, NCI-H69 cells were transfected with indicated plasmids and treated with MG132 for 2 hours before harvest and subjected to immunoprecipitation and immunoblot analysis as indicated. K, Immunoblot analysis of ASCL1 in ASCL1- and ASCL1–5SA-stable HEK293T cells treated with nocodazole for 24 hours. L, Luciferase analysis of Dll1-reporter and DLL3-reporter in wild-type and CDK2 knockdown NCI-H69 cells.

We noticed that ASCL1 protein displayed a mobility shift when co-expressed with CDK2 and CyclinA2 on SDS-PAGE (Fig. 2F). The band shift disappeared when cell lysates were treated with λ-PPase (Fig. 2G), indicating that the mobility shift was caused by phosphorylation modification. The in vitro kinase assay showed that CyclinA2 and CDK2 directly phosphorylated ASCL1 (Fig. 2H). PF-07104091 (PF-4091) is a specific CDK2 inhibitor. The phosphorylation modification of ASCL1 was inhibited by PF-4091 treatment (Supplementary Fig. S2B). Genetic inactivation of CDK2 also reduced the phosphorylation of ASCL1 (Supplementary Fig. S2C). Previous studies have demonstrated that CDK–Cyclin complexes tend to phosphorylate substrates with SP (serine proline) sites (26), and ASCL1 has five SP sites, S93, S190, S194, S207, and S223. We introduced point mutations at these five SP sites to mimic nonphosphorylated form (5SA) of ASCL1. As predicted from the potential CDK2 phosphorylation site, CDK2–CyclinA2 complex failed to promote the mobility shift of ASCL1–5SA (Fig. 2I). These results demonstrate that ASCL1 has multiple phosphorylation sites, and the CDK2–CyclinA2 complex directly phosphorylates ASCL1.

Next, we overexpressed WT-ASCL1 and ASCL1–5SA in NCI-H69 cells and examined their protein turnover in the presence of cycloheximide. The results showed that ASCL1–5SA was more stable than WT-ASCL1 (Supplementary Figs. S2D and S2E). Consistently, ASCL1–5SA had less ubiquitin modification compared with WT-ASCL1 (Fig. 2J; Supplementary Fig. S2F). Notably, CDK2–CyclinA2 complex failed to promote the ubiquitination of ASCL1–5SA (Supplementary Fig. S2G). Furthermore, we construct the HEK293T cell line stably expressing WT-ASCL1 or ASCL1–5SA proteins. The WT-ASCL1 stable cell line and the ASCL1–5SA stable cell line were synchronized with nocodazole. We found that ASCL1–5SA was not degraded during mitosis (Fig. 2K). Thus, these results suggest that ASCL1 is phosphorylated by CDK2–CyclinA2 complex and subjected to proteasome-mediated protein degradation.

Phosphorylation of transcription factors usually affects their localization or function, so we were wondering whether the phosphorylation of ASCL1 by the CDK2–CyclinA2 complex affects its nuclear localization. We overexpressed ASCL1, CDK2, and CyclinA2 in HeLa cells. Immunofluorescence assay revealed that the phosphorylation of ASCL1 by the CDK2–CyclinA2 complex did not alter its nuclear localization (Supplementary Fig. S2H). DLL1 and DLL3 are canonical downstream target genes of ASCL1. To explore whether phosphorylation affects the transcriptional activity of ASCL1, we built two ASCL1 luciferase reporter systems using mouse Dll1 and human DLL3 promoters which contain several E-box (CANNTG) binding sites (Supplementary Fig. S2I). Knockdown CDK2 or genetic inactivation of CDK2 or CDK2 inhibitor treatment decreased the transcriptional activity of ASCL1 (Fig. 2L; Supplementary Figs. S2J and S2K). qRT-PCR results indicated that CDK2 inhibition by PF-4091 reduced the expression of ASCL1 target genes (Supplementary Fig. S2L). In addition, ASCL1–5SA transcriptional activity was significantly reduced compared with WT-ASCL1 (Supplementary Fig. S2M). These results suggest that the CDK2–CyclinA2 complex promotes ASCL1 transcriptional activity by phosphorylating ASCL1.

HUWE1 mediates the degradation of ASCL1 during mitosis

HUWE1 belongs to the HECT-domain-containing ubiquitin ligase and is required for protein polyubiquitylation and proteasome-mediated protein degradation (27, 28). A previous study demonstrates that ASCL1 is the substrate of HUWE1 in mouse neural stem cells (29). Therefore, we examined HUWE1 expression in SCLC cells during mitosis. We found that HUWE1 protein levels was unchanged in H69 and H209, and upregulated in DMS53 during mitosis (Supplementary Fig. S3A). We tested whether HUWE1 is involved in cell-cycle-regulated ASCL1 degradation. To verify this hypothesis, we knocked down of HUWE1 in NCI-H69 cells. As predicted, ASCL1 protein levels were increased in HUWE1-knockdown NCI-H69 cells by two different siRNA targeting HUWE1 (Fig. 3A). Importantly, ASCL1 degradation was rescued in HUWE1-knockdown NCI-H69 cells when cells were synchronized in mitosis by nocodazole (Fig. 3B). This result was further confirmed in NCI-H209 cells and WT-ASCL1 stable HEK293T cell line (Fig. 3C; Supplementary Fig. S3B). In addition, we found that knockdown of Huwe1 prolonged the half-life of ASCL1 proteins in mouse SCLC cells (Supplementary Figs. S3C and S3D). These results indicate that HUWE1 is required for proteasome-mediated degradation of ASCL1 in SCLC cells.

Figure 3.

HUWE1 mediates the degradation of ASCL1 during mitosis. A, Immunoblot analysis of ASCL1 in NCI-H69 cells transfected with indicated siRNA. B and C, Immunoblot analysis of ASCL1 in HUWE1-knockdown NCI-H69 (B) and NCI-H209 (C) cells treated with nocodazole. D, Immunoprecipitation assay of ASCL1, ASCL1–5SA, and HUWE1 in HEK293T cells. E, Immunoprecipitation assay of ASCL1 and HUWE1 with or without overexpression of CDK2–CyclinA2 in HEK293T cells. F, Immunoprecipitation assay of ASCL1, ASCL1–5SA, and HUWE1 with or without overexpression of CDK2–CyclinA2 in HEK293T cells.

Next, we examined whether the phosphorylation modification affects the interaction between ASCL1 and HUWE1. The coimmunoprecipitation assay showed that HUWE1 interacted with WT-ASCL1, but not ASCL1–5SA (Fig. 3D). Furthermore, the CDK2–CyclinA2 complex promoted the interaction of WT-ASCL1 with HUWE1, but not ASCL1–5SA (Fig. 3E and F). However, knockdown of CDK2 did not rescue the expression of ASCL1 in mitosis (Supplementary Figs. S3E and S3F). These results suggest that the phosphorylation of ASCL1 by the CDK2–CyclinA2 complex promotes the interaction between ASCL1 and HUWE1 and further promotes proteasome-mediated ASCL1 degradation.

TCF3 protects ASCL1 from proteasome-mediated degradation and interacts with phosphorylated ASCL1

ASCL1 forms a heterodimer with E protein, such as TCF3, to execute its function (16). It has been shown that TCF3 binds to and protects ASCL1 from BMP2-induced degradation (30). We first explored whether TCF3 regulates the protein stability of ASCL1 in SCLC cells. Knockdown of Tcf3 in mouse SCLC cells by shRNA greatly reduced ASCL1 protein level, which is completely rescued by blocking proteasome-mediated protein degradation (Fig. 4A). In addition, cycloheximide chase assay showed that TCF3 overexpression significantly slowed down ASCL1 degradation (Fig. 4B). To further explore whether TCF3 could protect ASCL1 from degradation during mitosis, we overexpressed TCF3 in NCI-H69 cells and synchronized cells with nocodazole. The results showed that overexpression of TCF3 inhibited ASCL1 degradation during mitosis (Fig. 4C). Importantly, the association of ASCL1 with TCF3 was dramatically reduced during mitosis when NCI-H69 cells were arrested at G₂–M phase by nocodazole treatment (Fig. 4D). Thus, we speculated that the dissociation of ASCL1 from TCF3 exposes ASCL1 to HUWE1 and results in ubiquitin-modification and proteasome-mediated degradation of ASCL1.

Figure 4.

TCF3 promotes the stability and multiple sites phosphorylation of ASCL1. A, Immunoblot analysis of ASCL1 in Tcf3-knockdown mSCLC cells followed by MG132 treatment for 2 hours. B, Immunoblot analysis of ASCL1 in HEK293T transfected with indicated plasmids followed by CHX chase experiments. C, Immunoblot analysis of ASCL1 in NCI-H69 cells transfected with TCF3 and followed by nocodazole treatment for 24 hours before harvest. D, Immunoprecipitation assay of ASCL1 and TCF3 in mSCLC cells treated with nocodazole for 24 hours. E, HEK293T cells were transfected with indicated plasmids and cell lysates were treated with lambda phosphatase. F, HEK293T cells were transfected with different forms of ASCL1 with or without TCF3. G, HEK293T cells were transfected with indicated plasmids followed by BMS265246 treatment for 2 hours. H and I, Immunoprecipitation assay of ASCL1 variants and TCF3 in HEK293T cells.

We noticed that the overexpression of TCF3 resulted in a significant mobility shift of ASCL1. The mobility shift was abolished when cell lysates were treated with λ phosphatase, indicating that TCF3 promotes the phospho-specific modification of ASCL1(Fig. 4E). Previous study suggested that TCF3 stimulates CK2 kinase to phosphorylate ASCL1 at the S155 site (30). However, we found that TCF3 still promoted ASCL1-S155A phosphorylation (Fig. 4F). Of note, ASCL1 5SA did not have a mobility shift in the presence of TCF3 (Fig. 4F), indicating that TCF3 promotes ASCL1 phosphorylation at least through these five serine sites. We further introduced S155A mutation into the ASCL1–5SA, referred to ASCL1–6SA. We found that ASCL1–6SA acted the same as ASCL1–5SA with no mobility shift when co-expressed with TCF3 (Fig. 4F).

To investigate whether TCF3 could promote the phosphorylation of ASCL1 through the CDK2–CyclinA2 complex, we treated HEK293T cells with the CDK1/2 inhibitor BMS265246, which were overexpressed with ASCL1 and TCF3. The immunoblot analysis showed that BMS265246 partially inhibited the ASCL1 mobility shift produced by TCF3 (Fig. 4G), indicating that TCF3 promoted the ASCL1 phosphorylation partially through the CDK2–CyclinA2 complex. To further explore whether the phosphorylation status of ASCL1 will affect the interaction with TCF3, we performed coimmunoprecipitation assay. We found that phosphorylation-mimetic form of ASCL1 (5SE) enhanced the interaction between ASCL1 and TCF3, whereas 5SA mutant reduced it (Fig. 4H). 6SA mutation of ASCL1 totally abolished the association with TCF3 (Fig. 4I). These results suggest that TCF3 protects ASCL1 from proteasome-mediated degradation and preferentially binds with phosphorylated ASCL1.

The transcription of ASCL1 is inhibited by chemotherapy drug treatment

Most patients with SCLC are susceptible to chemotherapy drugs during the initial chemotherapy. Cisplatin and etoposide are the standard first-line treatment for metastatic SCLC. The drugs induce DNA damage and inhibit cell division. Given the mechanism of chemotherapy drugs, we speculated that EC may regulate ASCL1 protein expression. We performed subcutaneous tumor formation in nude mice with NCI-H209. When the tumor reached a mean diameter of around 0.5 cm, we gave them cisplatin and etoposide treatment as reported before (22). The IHC results showed that EC treatments significantly decreased the proportion of ASCL1-positive cells (Fig. 5A). In vitro assay also showed a sharp reduction of ASCL1 expression when treated NCI-H69 cells with EC (Fig. 5B). Consistently, the expression of ASCL1 target genes, such as DLL1, BCL2, and IGFBP5 was downregulated after drug treatment (Fig. 5C). However, blocking proteasome activity by MG132 treatment did not restore the protein level of ASCL1 (Fig. 5D). Next, we performed qRT-PCR assay to quantify the mRNA level of ASCL1 in NCI-H209 cells treated with EC. Surprisingly, we detected a markedly decreased mRNA level of ASCL1 (Fig. 5E). TSA has been shown to inhibited ASCL1 transcription through inducing Notch pathway activity in SCLC cells (24). However, we found that EC did not activate Notch pathway indicated by the expression of cleaved Notch1 and Notch target gene HES1 (Supplementary Figs. S4A and S4B). Because TCF3 stabilizes ASCL1 protein, we explored whether TCF3 degrades after EC treatment. However, TCF3 protein level was not affected by chemotherapy drugs (Fig. 5F). Intriguingly, we detected that the interaction between ASCL1 and TCF3 was enhanced after the drug treatment (Fig. 5G). These observations suggest that chemotherapy drugs modulate ASCL1 expression through transcriptional regulation but not posttranslational modifications.

Figure 5.

The transcription of ASCL1 is inhibited by chemotherapy drug treatment. A, IHC of ASCL1 expression (brown) using NCI-H209 subcutaneous tumors from nude mice. Scale bars, 50 μm. B, NCI-H69 cells were treated with EC for 24 hours and subjected to immunoblot analysis. C, Quantitative analysis of ASCL1 target genes mRNA level in NCI-H209 cells treated with EC. D, mSCLC cells were treated with EC followed by MG132 treatment and subjected to immunoblot analysis. E, Quantitative analysis of ASCL1 mRNA level in NCI-H209 cells treated with EC. F, Immunoblot analysis of TCF3 in mSCLC cells treated with EC. G, Immunoprecipitation assay of ASCL1 and TCF3 in mSCLC, NCI-H69, and NCI-H209 cells treated with EC. H, Knockdown of ASCL1 in NCI-H209 cells followed by EC treatment. Immunoblot analysis of cleaved PARP. I, Quantitative analysis of ASCL1 mRNA level in NCI-H209 cells. J, K, and L, The indicated NCI-H209 wild-type cells or ASCL1-knockdown cells were injected subcutaneously into nude mice. Representative images of tumors are shown in J. Growth curves of tumors were analyzed (K). Tumor weight measurement (L). Unpaired two-tailed Student t test between the indicated groups, n = 5 mice per group. ***, P < 0.001.

ASCL1 has been demonstrated as a proto-oncogene in SCLC (31). We next wanted to explore whether ASCL1 plays a role in the process of resistance to chemotherapy. We found that the depletion of ASCL1 sensitized NCI-H209 cells to cisplatin and etoposide treatment, which was indicated by the increased cleaved-PARP (Fig. 5H). To investigate whether the reduced ASCL1 expression affects the tumor growth, we generated the ASCL1 knockdown NCI-H209 cells and performed subcutaneous tumor formation in nude mice. The knockdown efficiency of ASCL1 was verified by qRT-PCR (Fig. 5I). We found that knockdown of ASCL1 significantly inhibited tumor growth (Fig. 5J–L). PTX (Taxol) inhibits the expression of ASCL1 through arresting SCLC cells in mitosis (Fig. 1B and C). To explore whether double inhibition PTX with EC would further suppress tumorigenesis, we performed xenograft tumor formation with nude mice (Supplementary Fig. S4C). Compared with EC treatment group, PTX in combination with EC group showed a better tumor regression (Supplementary Fig. S4D). However, this combination regimen has higher toxicity so it is not available for SCLC patients now (32).

Taken together, our results elucidate the regulatory mechanism of ASCL1 by the cell cycle and chemotherapy drugs (Fig. 6). Our finding provides a valuable theoretical basis for SCLC treatment.

Figure 6.

Diagram for the regulation of ASCL1 in SCLC by cell cycle. CDK2– complex phosphorylates ASCL1 and promotes ASCL1 degradation through HUWE1-mediates ubiquitination during mitosis. TCF3 protects ASCL1 from degradation and promotes the multisite phosphorylation of ASCL1. SCLC cells are more sensitive to chemotherapy drugs when the expression of ASCL1 is inhibited.

Discussion

ASCL1 is a proneural bHLH transcription factor, which is rarely mutated in tumors (33). Previous studies have found abnormal expression of proneural genes in different cancers such as colon cancer, lung cancer, thyroid cancer, brain cancer, and prostate cancer (34). ASCL1 is widely expressed in glioblastoma and malignant glioma (35, 36). ASCL1 is essential for maintaining and promoting the malignant characteristics of cancer stem cells (35). About 75% SCLC express ASCL1, which is required for the cell survival and growth in ASCL1-expressing SCLC (20). In our study, we elucidated a novel regulation of ASCL1 protein stability and transcriptional activity by cell-cycle-regulated kinase. We found that ASCL1 degrades during mitosis. CDK2–CyclinA2 complex directly phosphorylates ASCL1 and promotes its transcriptional activity in SCLC cells. Intriguingly, CDK1 and CyclinB also caused a mobility shift on ASCL1 and the mechanism of this finding needs to be addressed in future studies. We further showed that HUWE1 and TCF3 are involved in this regulation. Lack of ASCL1 makes SCLC cells more sensitive to EC treatment. Our research described here provides a path for the understanding and development of ASCL1-dependent cancer therapy.

SCLC is classified into four main subtypes by the expression of specific transcription factors. Among different molecular subtypes of SCLC, the ASCL1 positive subtype has the largest population (14, 37). Previous studies have shown that Ascl1-knockout mice cannot form SCLC tumors (15), indicating that ASCL1 is crucial for SCLC tumor formation. Elucidating the molecular mechanism of the ASCL1 regulation is critical for understanding SCLC tumorigenesis. As a proneural transcription factor, ASCL1’s activity should be precisely controlled. In this study, we found that phosphorylation of ASCL1 by cell-cycle-related kinase CDK2–CyclinA2 complex reduces its protein stability but increases the transcriptional activity. CDK2–CyclinA2 complex mainly functions in the S and G₂ phase and promotes the phosphorylation of ASCL1 (25). We showed that CDK2 and CyclinA2 directly interact with ASCL1 and promote its transcriptional activity in both human and mouse SCLC cell lines. CDK2 inhibitor decreases ASCL1 transcriptional activity. However, overexpression of phosphomutant ASCL1 in Xenopus embryo upregulated the differentiation targets such as Neural β-tubulin and Myt 1 (25). We speculate that the function of phosphorylated ASCL1 may exhibit a context dependent manner or CDK2 acts on specific ASCL1 phosphorylation sites in S-phase to promote transcriptional activity. One study reported that CDK2 inactivation destabilizes ASCL1 protein through an unknown mechanism in SCLC cells (38). CDK2 plays an important role in S and G₂ phase. But CDK1 compensates the function of CDK2 in Cdk2^−/- mice (39). The majority of cell types do not require CDK2 for their proliferation and survival (40). Thus, we speculate that other CDKs (such as CDK1) may be upregulated and compensate the role of CDK2 to phosphorylate ASCL1 in the absence of CDK2. ASCL1 has been shown to utilize an E3 ubiquitin ligase HUWE1-dependent mechanism for protein degradation (29, 41–43). HUWE1-mediated ubiquitination will be affected by the phosphorylation status of its substrates (44). Indeed, we showed that phosphorylation of ASCL1 by CDK2–CyclinA2 complex enhances the association with HUWE1. Meanwhile, TCF3 also preferentially binds to phosphorylated ASCL1 to exert transcriptional function. TCF3 promotes the phosphorylation modification of ASCL1 through CDK2 and cyclinA2 complex. TCF3 may compete with HUWE1 to bind to phosphorylated ASCL1 and protects ASCL1 from proteasome-mediated degradation during mitosis. Our analysis revealed that phosphorylated ASCL1 is labile form. Thus, dual regulation by TCF3 and HUWE1 precisely controls ASCL1 transcriptional activity. However, additional research will be needed to reveal how TCF3 dissociates from ASCL1 in mitosis.

SCLC is a well-known recalcitrant cancer. Etoposide combined with platinum agent was approved to treat SCLC over the past decades (14). Chemotherapy drugs, EC, impede DNA replication and cell division. We observed that EC treatment dramatically inhibit ASCL1 expression by repressing its transcription. One group analyzed the RNA-sequence data of 18 patients with relapsed SCLC and 80 treatment-naive patients with SCLC (37). They found that the relapsed samples have lower single-sample gene set enrichment analysis (ssGSEA) scores for ASCL1-driven gene expression (45). This clinical-related data strongly support our result that chemotherapy treatment reduces ASCL1 transcription. In our study, this inhibitory mechanism is distinct from cell-cycle regulation on ASCL1. The chemotherapy drug treatment does not activate Notch pathway to inhibit ASCL1 expression. Our data support that SCLC cells are sensitive to chemotherapy drugs when ASCL1 is downregulated. Recently, Kim, Chiho and colleagues reported that talazoparib (PARP inhibitor) in combination with JQ-1 led a more complete suppression of ASCL1 expression and xenograft tumors (46). These data imply that ASCL1 inhibition may improve the treatment efficacy in ASCL1 positive chemosensitive tumors. Therefore, targeting ASCL1 degradation in SCLC treatment is deserved to be further investigated.

Accumulating studies demonstrate that lineage plasticity causes drug resistance in prostate cancer and lung cancer (23, 47). It has been reported that MYC drives lineage plasticity to shift molecular subtypes in SCLC (48). We speculate that long-term drug stress may cause cell state change and finally change SCLC subtype properties. These findings highlight the importance of studying the dynamic regulations of these specific transcription factors in SCLC.

Our work reveals a new regulatory mechanism of ASCL1 by cell cycle and chemotherapy drugs in SCLC. Understanding the regulation of ASCL1 represents an important advance in the development of targeted therapy for the SCLC. It is possible that combination therapy targeting ASCL1 with chemotherapy drugs might benefit patients with SCLC.

Supplementary Material

Supplementary Figure S1

Supplementary Figure S1. ASCL1 protein level is regulated by the cell cycle.

Supplementary Figure S2

Supplementary Figure S2. Phosphorylation of ASCL1 reduces its protein stability by CDK2-CyclinA2.

Supplementary Figure S3

Supplementary Figure S3. HUWE1 mediates ASCL1 degradation during mitosis.

Supplementary Figure S4

Supplementary Figure S4. Notch pathway is not altered upon chemotherapy drug treatment.

Supplementary Table S1

Supplementary Table S1. Sequences for sgRNA, shRNA or siRNA.

Supplementary Table S2

Supplementary Table S2. Primer sequences for qRT-PCR.

Supplementary Table S3

Supplementary Table S3. Antibody information.

Authors' Disclosures

Y. Liu reports grants from National Natural Science Foundation of China, Natural Science Foundation of Zhejiang Provincial, China Postdoctoral Science Foundation, and National Key Research and Development Program of China during the conduct of the study. H. Song reports grants from National Natural Science Foundation of China, Natural Science Foundation of Zhejiang Provincial, and National Key Research and Development Program of China during the conduct of the study. No disclosures were reported by the other authors.

Authors' Contributions

Y. Liu: Resources, data curation, software, formal analysis, funding acquisition, validation, investigation, methodology, writing–original draft, project administration, writing–review and editing. Q. Wu: Resources, validation, investigation. B. Jiang: Validation, investigation, project administration. T. Hou: Investigation. C. Wu: Investigation, project administration. M. Wu: Resources, supervision, funding acquisition, methodology, writing–review and editing. H. Song: Resources, formal analysis, supervision, funding acquisition, validation, investigation, project administration, writing–review and editing.