Mechanistic Exploration of Cancer Stem Cell Marker Voltage-Dependent Calcium Channel α2δ1 Subunit-mediated Chemotherapy Resistance in Small-Cell Lung Cancer

Small-cell lung cancer (SCLC) is an aggressive cancer of neuroendocrine origin with poor prognosis, accounting for approximately 15% of all lung cancer cases. The majority of SCLC patients have extensive stage disease (ES) with metastasis beyond one hemithorax at the time of diagnosis (1). Despite the robust response to initial chemotherapy and radiotherapy, most patients progress within 3 months and develop resistance to the initial treatment. Less than 30% of patients with limited disease and 1%–2% of patients with extensive disease survive to 5 years (2).

Putative cancer stem cells (CSC) have been identified in many human cancer types, including SCLC (3). CSCs are a small subset of cells that sustain tumor growth and remain in patients after conventional antitumor therapy has been completed (4). CSCs are also reported to be responsible for tumor initiation, maintenance, and spreading and are characterized by capacity for self-renewal, asymmetric cell division, slow division kinetics, increased capacity for invasion, metastasis, tumor formation and proliferation, and resistance to conventional chemotherapy and radiotherapy. They can be identified by a variety of cell markers (5–8). CSC populations are commonly isolated and enriched using cell surface markers such as CD133, CD44, CD87, and ALDH1. CD133 is a transmembrane glycoprotein and is the most widely used stem cell markers to identify CSCs from various tissue sources such as neural or brain tumors, renal epithelial, or kidney cancer and lung cancer (3, 9–12). SCLC contains identifiable populations of stem cells (13), which are thought to be able to evade conventional therapy and are implicated in chemotherapy resistance and high recurrence rate of SCLC, both of which are major challenges in SCLC. In a preliminary study, it was demonstrated that a small subset of cells in SCLC expresses CD133 and possess stem cell–like properties.

Freshly dissociated human SCLC contains CD133-positive cells, which can generate long-term lung tumor spheres in vitro that can both differentiate and preferentially form tumors in vivo (3). Sarvi and colleagues reported that level of CD133 was elevated in SCLC tumors following chemotherapy (14). However, the evidence for CD133 as an effective CSC marker in lung cancer remains contradictory (15, 16). Kubo and colleagues studied CD133 and CD87 as CSC markers in a panel of six SCLC cell lines and concluded that CD133 and CD87 were inadequate CSC markers in SCLC (17). Meng and colleagues found that CD133 status in A549 and H446 cell lines was not significantly related to proliferative capacity, invasiveness, drug resistance, or tumorigenic ability in xenograft models (16). Because CSCs are phenotypically heterogeneous, no marker or set of markers have been found to identify CSCs in solid tumors in general nor for specific tumor entities. Therefore, new biomarkers are necessary for further identification of CSCs.

Recent research has identified the voltage-dependent calcium channel α2δ1 subunit, which is part of a voltage-gated calcium channel complex, as a promising potential marker for CSCs. Zhao and colleagues (18) studied resistant primary hepatocellular carcinoma (HCC) and found the α2δ1⁺ cells presented as CSCs. They also found that α2δ1 was a functional marker for prediction of HCC recurrence. The functional part was identified and validated as isoform 5 of α2δ1, and its monoclonal antibody (mAb) 1B50-1, is a potential anti-HCC agent. Whether α2δ1 subunit plays the same role in SCLC CSCs warrants further exploration.

We hypothesized that α2δ1 might be an excellent marker of CSCs, comparable with CD133, in SCLC. To test this hypothesis, we investigated the expression levels of CD133 and α2δ1 using three SCLC cell lines and five PDX models. Meanwhile, we examined whether α2δ1 might be effective for selecting cancer stem-like cells from SCLC. We also explored the relationship between α2δ1⁺ CSCs and mechanisms of chemotherapy resistance in SCLC. Deep understanding of this population may be helpful in addressing chemoresistance and seeking novel strategies to overcome chemoresistance in SCLC patients.

Expression of stem cell-like markers was analyzed in NCI-H1048, NCI-H69, and NCI-H209 SCLC cell lines. All the cell lines were from BeiGene (Beijing) Co. Ltd., and were cultured in RPMI1640 supplemented with 10% FBS and 1% penicillin/streptomycin at 37°C under 5% CO₂. The NOD/SCID female rats and Balb/c female rats (aged 4–6 weeks, weighing 18–21 g) used in vivo experiments were purchased from Huafu Kang Experimental Animal Co., Ltd. All animals were fed in the SPF class room by the Peking University Clinical Cancer Hospital Animal Center.

Five SCLC patient-derived xenografts (PDX) including Bclu80, Bclu62, Bclu82, Bclu44, and Bclu77 were used to explore the expression of CSC marker and role of drugs. Five PDX models kept high fidelity with the primary patients' tumors whether histopathology [hematoxylin and eosin (H&E) staining], immunohistochemistry (IHC) staining (common SCLC markers, CD56, TTF1, CgA, and Syn) or therapy response to standard regimens (Supplementary Table S1). Histologic analysis was performed by two experienced pathologists (with more than 5 years of experience in lung cancer) who were blinded to the clinical data. Histopathology and IHC staining of five PDX models both showed typical characteristics of SCLC.

All animal experiments were performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals with protocols approved by the Animal Care and Use Committee at Peking University Cancer Hospital.

Cells were collected and resuspended at 1 × 10⁶ cells/mL of PBS buffer. Fluorescence-labeled mAbs were added at the concentrations recommended by the manufacturer. After washing, the labeled cells were analyzed and sorted using BD Accuri C6 and FACS Aria flow cytometer (Becton Dickinson), respectively. The antibodies used were APC/PE–conjugated mouse anti-human CD133 and corresponding isotype IgG. Detailed information of antibodies is listed in Supplementary Table S2. The 1B50-1 antibody and isotype IgG were supplied by Laboratory of Carcinogenesis and Translational Research (Ministry of Education), Department of Cell Biology, Peking University Cancer Hospital and Institute (Beijing, China). 1B50-1 was directly labeled with PE-Cy5 or fluorescein using the respective Lightning conjugation kits following the manufacturer's protocol (Innova Biosciences Ltd.; ref. 20).

After counting, cells were mixed with DMEM/F12 medium (Invitrogen) supplemented with B27 (Invitrogen), 20 ng/mL epidermal growth factor, 20 ng/mL basic fibroblast growth factor (PeproTech), 10 ng/mL hepatocyte growth factor (PeproTech), and 1% methylcellulose (Sigma), and were cultured in semisuspension in Ultra Low Attachment 96-well plates (Corning Incorporated Life Sciences; ref. 20). The number of spheres was counted and recorded under a stereomicroscope (Olympus) after 2- to 3-week incubation.

Cells purified by fluorescence-activated cell sorting (FACS) were suspended in PBS and mixed 1:1 with Matrigel (BD Biosciences). Suspensions of 10²–10³ cells were transplanted subcutaneously bilaterally in the back of NOD/SCID mice to investigate the capacity of tumorigenicity. The tumor was observed and measured twice weekly.

Reverse transcription-PCR (RT-PCR) was applied to detect the expression of stem cell–related transcription factors (SOX2, OCT4, and Nanog) and drug resistance–related genes (MDR, ABCG2) in α2δ1/CD133–positive and negative cells. The expression of CACNA2D1 (α2δ1-related gene) in PDX model was also detected by RT-PCR. The FACS-purified cells were lysed by QIAzol Lysis Reagent and total RNA was extracted using an RNeasy Mini Kit (Qiagen) according to the manufacturer's protocol. Tumor tissue from PDX model was lysed into a single-cell suspension and RNA was extracted using TRIzol and chloroform. cDNA was synthesized using EasyScript First Strand cDNA Synthesis SuperMix (TransGen). The quantitative PCR reaction was carried out in a Stratagene Mx3000P system and data were analyzed using the System detection software (MxPro-Mx3000P software, Agilent Technologies). The primers of each gene are listed in Supplementary Table S3.

Cells were lysed using cell lysis buffer (Cell Signaling Technology) and quantified by a bicinchoninic acid assay (Applygen). Lysates were resolved on 12% SDS-PAGE gels and electroblotted onto nitrocellulose membranes. Primary antibodies were added and incubated overnight at 4°C followed by horseradish peroxidase–labeled anti-rabbit or anti-mouse secondary antibodies (Cell Signaling Technology) for 2 hours at room temperature. Bands were visualized using enhanced chemiluminescence and photographed using a Fujifilm Dark Box II. Primary antibodies included ERK1/2, pERK1/2, pAKT, pS6, SOX2, and β-actin, etc. Detailed information of antibodies applied are listed in Supplementary Table S4.

To examine the sensitivity of the different cell populations to chemotherapy, we treated H1048 cells with different concentrations of etoposide (the standard chemotherapy regimen for SCLC) for 72 hours. Following exposure to the drug, we analyzed expression of α2δ1 and CD133 by flow cytometry. We implanted Balb/c mice with H1048 cells. After the tumors grew to a volume of about 200 mm³, we administered standard chemotherapy for SCLC (etoposide and cis-platinum, EP) for 3 cycles. We obtained the residual tumors and analyzed α2δ1 and CD133 expression in EP-treated mice as compared with untreated mice by flow cytometry.

For detailed methods regarding the establishment of PDX models, please refer Supplementary Data.

Animals were treated by intraperitoneal injection with etoposide and cisplatin (EP) [4 mg/kg cisplatin dissolved in 0.9% saline solution on day 1, and 12 mg/kg etoposide (Sigma) dissolved in 0.9% saline solution on days 1–3], or the corresponding vehicle only, 10 days/cycle for a total of 3 cycles. Individual body weight and tumor volume were recorded twice weekly, with mice monitored daily for clinical signs of toxicity for the duration of the study. When the tumor volume reached 2,000 mm³, or when the tumor was ulcerated, mice were euthanized using carbon dioxide.

Chemotherapeutic response was evaluated as the percentage of tumor growth inhibition (% TGI), calculated using the equation “100 − (the mean RTV of the treated tumor/the mean RTV in the control group at the time of sacrifice × 100)”, where RTV (relative tumor volume) was defined as Vx/V1, in which Vx was the volume in mm³ at a given time and V1 was the volume at the start of treatment. Mean RTV and SD were calculated for each group (19).

To test the therapeutic effect of 1B50-1, we treated PDX models Bclu80 and Bclu62 with 3 cycles of EP regimes combined or sequenced with 1B50-1 antibody (800 μg/mouse, once daily, i.p., until disease progression). For PDXs and corresponding clinical data, refer to Supplementary Table S1.

Total RNA was extracted from α2δ1⁺ and α2δ1⁻ cells sorted by FACS using All Prep DNA/RNA Mini Kit (Qiagen) according to the manufacturer's instructions. Sequencing libraries were generated using NEBNext Ultra Directional RNA Library Prep Kit for Illumina (NEB) following manufacturer's recommendations. Briefly, mRNA was purified from total RNA using poly-T oligo-attached magnetic beads. Fragmentation was carried out using divalent cations under elevated temperature in NEBNext First Strand Synthesis Reaction Buffer (5×). First-strand cDNA was synthesized using random hexamer primer and M-MuLV Reverse Transcriptase (RNaseH-). Second strand cDNA synthesis was subsequently performed using DNA Polymerase I and RNase H. The library fragments were purified with AMPure XP system (Beckman Coulter). Then PCR was performed with Phusion High-Fidelity DNA polymerase, Universal PCR primers and Index (X) Primer. At last, products were purified (AMPure XP system) and library quality was assessed on the Agilent Bioanalyzer 2100 system. The cDNA library was sequenced on a HiSeq 2500 platform (Illumina) using PE125 sequencing strategy.

Raw data of fastq format were firstly processed through in-house perl scripts. In this step, clean data (clean reads) were obtained by removing reads containing adapter, reads containing ploy-N and low quality reads from raw data. Reference genome and gene model annotation files were downloaded from UCSC (genome.ucsc.edu/) directly. Index of the reference genome was built using Bowtie v2.0.6 (20) and paired-end clean reads were aligned to the reference genome using TopHat v2.0.9 (21). HTSeq v0.6.1 (22), which considers the effect of sequencing depth and gene length for the reads count at the same time, was used to count the reads numbers mapped to each gene. And then RPKM (Reads Per Kilobase of exon model per Million mapped reads) of each gene was calculated on the basis of the length of the gene and reads count mapped to this gene. Prior to differential gene expression analysis, for each sequenced library, the read counts were adjusted by edgeR program package through one scaling normalized factor. Differential expression analysis of two conditions was performed using the DEGSeq R package (1.12.0; refs. 23, 24). The P values were adjusted using the Benjamini–Hochberg method. Corrected P value of 0.05 and log₂ (fold change) of 1.5 were set as the threshold for significantly differential expression.

All statistical analyses were performed with the SPSS statistical software, version 19.0 (IBM Corp.). Comparisons between groups were performed by paired Student t test, independent t test, or one-way ANOVA. Bonferroni correction or Dunnett T3 method was used for the multiple comparisons. The statistical significance level was defined as two-sided P < 0.05.

The expression of CD133 and α2δ1 by flow cytometry analysis were different in several SCLC cell lines. The expression of CD133 in NCI-H1048, NCI-H69, and NCI-H209 was 41.10% ± 4.63%, 93.22% ± 3.42%, and 91.05% ± 3.68%, respectively, while the α2δ1 expression was 1.40% ± 0.26%, 12.08% ± 2.23%, and 4.20% ± 1.85%, respectively (Fig. 1A). A relationship between α2δ1⁺ and CD133⁺ was observed in H1048. The expression of CD133 in H1048 cells was significantly higher than α2δ1 (P < 0.01; Fig. 1B), and CD133⁺ population is nearly 100% positive in α2δ1⁺ cells, while only 2.67% α2δ1⁺ cells occur in CD133⁺ cells. (Fig. 1C). Total α2δ1⁺ cells are a particular subpopulation in CD133⁺ cells, which regards as CD133⁺/α2δ1⁺ cells in H1048. Confocal laser scanning microscopy showed that CD133 was expressed on the membrane, and α2δ1⁺ demonstrated granula membrane expression (Fig. 1D).

α2δ1^+/− and CD133^+/− populations of H1048 cell line were identified and sorted by FACS and cultivated in semisolid media without serum to evaluate self-renewal capacity. With FBS-free medium, we performed the clone formation ability of α2δ1⁺ and CD133⁺ cells compared with control cells. Results suggested both α2δ1⁺ and CD133⁺ cells demonstrated higher sphere-forming efficiency compared with α2δ1⁻ and CD133⁻ populations as well as parental H1048 cells, respectively (Fig. 2A and 2B). The sphere-forming ability of α2δ1⁺ cells were even stronger than CD133⁺ cells (P = 0.013). What's more, cells separated from α2δ1⁺ or CD133⁺ spheres could be passaged and the efficiency of sphere formation elevated with passage (Fig. 2C). For demonstrating in vivo, FACS-sorted α2δ1^+/− and CD133^+/− cells of H1048 were bilaterally inoculated into the backs of NOD/SCID mice with different magnitude (10² or 10³) to observe the tumorigenic ability (Fig. 2D). Double positive (α2δ1⁺CD133⁺) cells exhibited higher frequency of tumorigenicity compared with α2δ1⁻CD133⁺ and α2δ1⁻CD133⁻ cells, respectively (Supplementary Table S5). Moreover, α2δ1⁺ and CD133⁺ cells had higher tumor formation rates compared with α2δ1⁻ and CD133⁻ population, respectively, in H1048, H69, Bclu80 PDX (Supplementary Table S6). Serial transplantation assay was performed into mice for validating self-renewal in vivo. Both α2δ1⁺ and CD133⁺ cells constantly obtained the ability of self-renewal when generating. However, α2δ1⁺ cells exhibited no significantly higher growth rate than only CD133⁺ cells at the same concentration (Supplementary Table S7), and continuously feed for 80 days, tumor volumes were measured as shown in Fig. 2E. A series of passaged paraffin-embedded sections were examined by H&E examination. The morphology of cells remained consistent between different passages and also with parent sample (Fig. 2F). α2δ1⁺ cells also presented an elevated self-renewal capability in H69 cell lines both in vitro and in vivo (Supplementary Fig. S1). These data indicated α2δ1⁺/CD133⁺ cells contribute to higher tumorigenic ability in SCLC.

When cultured in medium with complete serum (10% FBS), the purified α2δ1⁺ and CD133⁺ cells of H1048 cell line showed differentiation ability. After two weeks of culture, flow cytometry analysis showed that the proportion of α2δ1⁺ and CD133⁺ cells returned from 98.2% and 97.8% to baseline level 1.1% and 43.2%, respectively (Fig. 3A). Moreover, FACS-sorted purified α2δ1⁺/CD133⁺ cells from the first-generation tumors formed by α2δ1⁺/CD133⁺ cells were transplanted into NOD/SCID mice again to examine the serial transplantation capacity in vivo. The proportion of α2δ1⁺ and CD133⁺ cells in second-generation PDXs returned from 95.74% ± 1.44% and 97.49% ± 1.24% to the parental level 1.22% ± 0.27% and 44.9% ± 2.2% (Fig. 3B). This phenomenon was also observed in serial passage of α2δ1⁺ Bclu80 PDX models (Supplementary Fig. S2). It demonstrated the differentiation properties of α2δ1⁺ and CD133⁺ cells were conserved both in vitro and in vivo.

To verify the stem cell–like properties of α2δ1⁺ and CD133⁺ cells and investigate the relationship with drug resistance, RT-PCR was applied to detect the expression of stem cell–related transcription factor (SOX2, OCT4, and Nanog) and drug resistance–related genes (MDR, ABCG2) in FACS-sorted α2δ1^+/− and CD133^+/− H1048 cells. Results showed that all of these genes were more highly expressed in α2δ1⁺ and CD133⁺ cells compared with their corresponding negative cells. Notably, the expression levels were even higher in α2δ1⁺ cells than in CD133⁺ cells, especially for OCT4, MDR1, and ABCG2 (Fig. 3C).

Next, we evaluated the response of corresponding PDXs with different expression of α2δ1 and CD133 to chemotherapy. Tissue clumps of SCLC patients from bronchoscopy biopsy were transplanted into NOD/SCID mice to establish PDXs. Preliminary experiments showed that the PDXs were highly consistent with primary tumor of patients with respect to morphologic, pathologic, genomic, and clinical characteristics. Pathologic and genetic characteristics were highly conserved throughout passaging (25).

After successful PDX establishment, the PDX tumors were disaggregated into single-cell suspension and analyzed for α2δ1 and CD133 expression by flow cytometry. Results showed that α2δ1 and CD133 expression varied between individual PDXs. Of note, Bclu80 model presented significantly higher expression of α2δ1 compared with Bclu62, Bclu82, Bclu44, and Bclu77, while expression of CD133 was relatively lower (Fig. 3D). The expression ratio of α2δ1/CD133 showed similar trend with the expression of α2δ1 in corresponding PDXs (Fig. 3E). Moreover, the expression of α2δ1-related genes (CACNA2D1) examined by RT-PCR was also consistent with the expression of α2δ1 protein determined by flow cytometry (Fig. 3F). We also explored the stem cell properties of α2δ1⁺ cells in PDX models. α2δ1⁺ populations showed high efficiency of sphere-forming and tumorigenicity compared with α2δ1⁻ cells in the Bclu80 model. (Supplementary Fig. S4; Supplementary Table S6).

In addition, PDXs with different expressions of stem cell markers showed diverse responses to chemotherapy. The models with relatively low α2δ1 expression or a low α2δ1/CD133 ratio (Bclu62, Bclu82, and Bclu77) showed good efficacy to EP chemotherapy with best response of partial response (PR) or even complete response (CR). However, Bclu80, which has high expression of α2δ1 and a high α2δ1/CD133 ratio, was not sensitive to EP chemotherapy with a best response of stable disease (SD). Accordingly, the response of PDXs to chemotherapy was consistent with clinical response of patients. Surprisingly, the Bclu44 model, which displayed low α2δ1 expression and a low α2δ1/CD133 ratio, showed only SD or limited remission (minimal PR) in PDXs. However, the corresponding patient acquired a clinic response of PR to chemotherapy (Supplementary Tables S1 and S8; Supplementary Fig. S3). CD133 showed different expression level in five PDXs. However, no correlation could be observed between CD133 expression and clinical response.

To investigate the effects of chemotherapy on expression of stem-like cells, H1048 cells were treated with different concentrations of etoposide for more than 72 hours, and α2δ1 and CD133 were evaluated by flow cytometry. The proportion of α2δ1⁺ and CD133⁺ cells increased after treating with etoposide and the α2δ1⁺ population was enriched even more after treatment (Fig. 4A). Similarly, the proportion of α2δ1⁺ and CD133⁺ cells significantly increased in tumors treated after 3 cycles of EP chemotherapy as compared with those in control group in H1048-transplanted mice (Fig. 4B). Analogous with the in vitro results, the α2δ1⁺ population showed a more significant enrichment after treatment than the CD133⁺ population.

Tumors from Bclu62 and Bclu82 PDX models were sensitive to chemotherapy with a best response of PR. The proportion of α2δ1⁺ cells in the Bclu62 and Bclu82 PDX tumors after 3-cycle chemotherapy of EP regimes increased significantly. However, CD133 expression remained stable (Fig. 4C and D). The Bclu80 PDX showed a response of SD with minimal shrinkage, and the proportion of α2δ1⁺ and CD133⁺ population did not significantly change before and after chemotherapy (Fig. 4E). Similarly, nonsensitive Bclu44 PDX showed the same trend as Bclu80 (Fig. 4F). Chemotherapy induced the enrichment of α2δ1⁺ population rather than CD133⁺ cells.

1B50-1 antibody is a specific mAb against the α2δ1 subunit. We treated Bclu80 PDX with both concurrent therapy (chemo and 1B50-1 antibody) and subsequent therapy (chemo followed by 1B50-1 antibody). The group of combined treatment with EP regimes and 1B50-1 antibody in concurrent therapy showed an improved best response (tumor shrunk 75.5% compared with baseline) than the group treated with chemotherapy alone (tumor shrunk 49.7%), and the tumors in group of 1B50-1 antibody also exhibited slow growth compared with those in control group (Fig. 5A). Similar results were observed in the Bclu62 PDX model (Fig. 5B). In the alignment of sequential treatment of Bclu80 PDX, all animals received 3 cycles of EP regimes and then were grouped randomly in 1B50-1 antibody and control group. The antibody-treated group showed significantly slower tumor growth compared with control group (Fig. 5C). Together, these data indicated that 1B50-1 antibody is effective in slowing down tumor growth whether combined with chemotherapy or alone, although the superiority of single-agent therapy was not obvious as compared with chemotherapy. We also explored the proportion of α2δ1⁺ population in different treatment groups. The groups treated with 1B50-1 antibody showed relatively low proportion of α2δ1⁺ cells compared with control and EP chemotherapy groups (Fig. 5D).

After confirmed the expression ratio of α2δ1 related to the chemosensitivity and chemotherapy induced the enrichment of α2δ1⁺ cells in PDX models, we then evaluated the sensitivity of α2δ1⁺ cells to chemotherapy. The sensitivity of parental cells versus the passage cells digested from spheres of α2δ1⁺ cells to conventional chemotherapy regimen (etoposide and cisplatin) was examined in vitro. Results showed that the IC₅₀ for sphere cells was higher compared with parental cells treated with both etoposide (0.546 μmol/L vs. 0.062 μmol/L; Fig. 6A, left) and cisplatin (11.3 μmol/L vs. 1.8 μmol/L; Fig. 6A, right) after 72-hour plate culturing, which suggested sphere cells exhibited robust resistance to conventional chemotherapy. In addition, the sphere formation ability of parental and α2δ1^+/− cells with cisplatin treatment was tested, and α2δ1⁺ cells showed stronger sphere formation ability and high risk of resistance (Fig. 6B).

α2δ1⁺ cells exhibited the properties of CSC and chemoresistance. RNA-seq was used to explore the different gene expression profile of α2δ1⁺ and α2δ1⁻ populations FACS-sorted from parental H1048 cells (Supplementary Table S9). There were 28 genes (with at least 3.0-fold changes) concurrently upregulated and 8 genes downregulated in α2δ1⁺ cells compared with α2δ1⁻ population (Supplementary Fig. S5A and S5B). By mapping all of these genes to the KEGG pathway database, the involved signaling pathways included the MAPK, PI3K-AKT, cell adhesion molecules (CAM), and p53 pathways, etc. (Supplementary Fig. S5C).

Considering the transcription test results and previous literature, we hypothesized that the MAPK pathway activated in α2δ1⁺ population and that this may contribute to the stem cell properties and chemoresistance. The crucial proteins involved in MAPK, PI3K–AKT pathways, and SOX2 protein were detected in FACS-sorted α2δ1^+/− H1048 cells using Western blotting. The proteins of pERK1/2 and SOX2 were both over expressed in α2δ1⁺ population compared with α2δ1⁻ cells (Fig. 6C).

We further investigated the effect of overexpression of α2δ1 on the expression of stem cell and drug resistance–related proteins and genes in MAPK and apoptosis-related pathways. As shown in the Fig. 6D, after overexpression, the upregulated proteins included BMI1, Nanog, OCT4, p-ERK, BCL2, combined with the downregulation of proapoptosis proteins BAX and cleaved caspase-3. For gene expression, compared with the vector, α2δ1 overexpression upregulated stem cell–related genes (BMI1, EPCAM, NANOG, SOX2, KLF4, and CTNNB) and drug resistance genes (ABCG2 and MDR1; Fig. 6E).

MAPK pathway involved the properties of stem cell and drug resistance of α2δ1⁺ cells. We further investigated the role of ERK inhibitor alone or in combination with 1B50-1 antibody in vitro. As shown in Fig. 6F, when α2δ1⁺ cells were treated with single neutralized 1B50-1 antibody, single-agent ERK inhibitor (SCH772984) and combined 1B50-1 antibody/ERK inhibitor, the sphere formation rate was all decreased. What is more, combined SCH772984 and 1B50-1 antibody could inhibit the growth of α2δ1⁺ cells in the greatest extent. This suggested that combined ERK inhibitor and 1B50-1 treatment may conquer the α2δ1⁺ sphere formation ability.

This study is the first that we are aware of to evaluate expression of α2δ1 in SCLC cells and PDXs models by FACS, to verify the stem cell–like properties of α2δ1⁺ cells, and to identify the difference of α2δ1 and CD133 expression between chemosensitive and resistant PDXs models. Our data suggest that α2δ1⁺ SCLC cells demonstrate CSC properties, and might contribute to chemotherapy resistance. Furthermore, we have demonstrated that use of a 1B50-1 antibody could help overcome chemoresistance and delay relapse of SCLC in PDX models.

The target of 1B50-1 has been previously identified and validated as isoform 5 of α2δ1, a subunit of voltage-gated calcium channel complexes (26). Currently, little is known about the expression pattern and specific function of isoform 5 of α2δ1. Our study found that α2δ1⁺ cells from H1048 had higher sphere-forming efficiency in vitro, and were highly tumorigenic in vivo. The α2δ1⁺ cells also had a high differentiation potential when cultured with complete serum, suggesting that α2δ1 marked stem cells from the H1048 cell line exhibited stem cell properties. Therefore, α2δ1 identifies a highly tumorigenic and invasive subpopulation of SCLC. OCT4, Nanog, and SOX2 are three basic transcription factors that are expressed in both CSC-like cells and embryonic stem cells (27). In our study, stem cell–related transcription factors (SOX2, OCT4, and Nanog) and drug resistance–related genes (MDR and ABCG2) were more highly expressed in α2δ1⁺ and CD133⁺ cells, especially in α2δ1⁺ cells. Also, compared with CD133⁺ cells, α2δ1⁺CD133⁺ population exhibited not only higher sphere forming ability in vitro, but also relatively higher tumorigenicity, growth rate and proliferation property in vivo. These indicated that α2δ1⁺ cells exhibit stem cell–like properties in vitro and in vivo, and α2δ1 could be a used as a marker for CSC screening in SCLC. At the same time, α2δ1 overexpression may attribute to the resistance of chemotherapy of SCLC.

Many studies have demonstrated that PDX models can be used as surrogates for clinical biopsies of lung cancer patients in research (28–31), including our former study (25). Therefore, we utilized PDX models and flow cytometry to analyze the two markers. We found that CD133 was overexpressed in most PDXs and that α2δ1 expression varied in individual PDXs. Furthermore, we treated H1048 cells and PDXs with chemotherapy, and found out that α2δ1/CD133–positive cells in H1048 cells significantly increased after treatment compared with control group, and α2δ1-positive cells increased even more. We observed the same trend in the tumors derived from H1048 cells in mice as well after they were treated with chemotherapy. However, in subgroup of PDXs with partial response (PR) to chemotherapy, significantly increased α2δ1 and stable CD133 expression was observed after chemotherapy in PDXs. Contrarily, α2δ1 and CD133 expression stay the same before and after chemotherapy in nonresponsive PDXs.

The enrichment of α2δ1 and CD133 subpopulations after chemotherapy from H1048 cell line and PDXs were not exactly the same. The possible reasons are as follows: first, H1048 cell line–based xenografts grow primarily as homogeneous masses of cancer cells with minimal stromal infiltration; therefore, they might not recapitulate the human tumor architecture and interactions between stromal components and patient tumor cells (32). Second, PDXs should be better models for representing the condition of patient's tumor. In vivo studies are always more convincing than in vitro studies because in vitro experiments fail to replicate the complex cellular conditions of an organism and sometimes may lead to results that do not correspond to the circumstances occurring around a living organism (33). Keeping these limitations in mind, our data from in vitro, in vivo, and PDXs consistently showed that α2δ1⁺ subpopulation might contribute to resistance and recurrence. What is more, higher α2δ1 expression or a higher ratio of α2δ1/CD133 were related to chemoresistance. This suggests that α2δ1 overexpression should be an independent factor for chemoresistance in vivo.

We also performed transcriptome sequencing to explore the possible mechanisms of α2δ1-mediated chemoresistance (Supplementary Fig. S5). We found an enrichment of genes related to MAPK, PI3K-AKT and CAM signaling pathways. Combined analysis with previous literature (20), we speculated MAPK signaling pathway might be more important. We explored the expression of crucial proteins involved in MAPK, PI3K-AKT pathways, and SOX2 protein, and found overexpression of phosphorylated ERK from MAPK pathway in α2δ1⁺ H1048 cells. ERK1/2 are viewed as canonical MAPKs with cytoplasmic and nuclear substrates involved in signal transduction and the regulation of transcription. These two kinases play redundant roles in proliferation and differentiation. Emerging evidence suggests that in addition to having a possible role in differentiation and proliferation, activation of the RAF/MEK/ERK pathway can promote cell invasion and metastasis in SCLC cells (34–37). There are no studies currently investigating the relationship between the voltage-dependent calcium channel α2δ1 and stem cells in SCLC. However, recently Ciccarelli and colleagues reported that the MEK/ERK pathway plays a prominent role in maintaining the stem-like phenotype of rhabdomyosarcoma cells, their survival and their innate radioresistance (38). In combination with our data demonstrating overexpression of phosphorylated ERK in α2δ1⁺ SCLC cells, this suggests that the role of the MEK/ERK pathway in SCLC chemoresistance should be further investigated.

Zhao and colleagues generated the mAb, 1B50-1, which had a therapeutic effect on hepatic cell carcinoma (18). Our study is the first to treat SCLC PDXs with 1B50-1 antibody, and we found that originally chemoresistant PDX80 developed PR after combination treatment with 1B50-1 antibody and chemotherapy. However, no superiority of single-agent 1B50-1 treatment was seen compared with chemotherapy alone or combined therapy. This may suggest that stem cells only make up a small portion of tumor cells in SCLC that could be responsive to therapeutic agent targeted at stem cell markers such as α2δ1 instead of standard chemotherapy.

The limitations of current study are that first this is a small sample size, and second it would be better if we also included the primary patients derived cell lines for in vitro study. These results should be further validated in additional samples and in the clinical setting.

In conclusion, α2δ1 subunit–positive SCLC cells have CSC properties, and are related to chemotherapy resistance. The ERK protein in MAPK pathway was highly expressed in the α2δ1⁺ H1048 cell line, and might be the critical molecular for resistance mediated by the α2δ1 subunit. Use of a specific antibody of 1B50-1 to α2δ1 subunit could improve response to chemotherapy and delay relapse, whether combined with chemotherapy or as sequential therapy.

No potential conflicts of interest were disclosed.

Conception and design: J. Yu, J. Wang

Development of methodology: J. Yu, S. Wang, Z. Wang, J. Wang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Yu, S. Wang, W. Zhao, H. Chen, Y. Tian, D. Wang, J. Zhao, T. An, M. Wu, J. Wang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J. Yu, S. Wang, W. Zhao, Z. Wang, Y. Tian, D. Wang, J. Wang

Writing, review, and/or revision of the manuscript: J. Yu, S. Wang, W. Zhao, H. Chen, T. An, J. Wang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J. Yu, S. Wang, W. Zhao, J. Duan, T. An, M. Wu, J. Wang

Study supervision: J. Yu, H. Bai, J. Wang

Other (submission of the manuscript): S. Wang

We thank Zhiqian Zhang and Haibo Han from the Laboratory of Carcinogenesis and Translational Research (Ministry of Education), Department of Cell Biology for the supply of antibody, experiment assistance and instruction, and Huirong Ding, Zhihua Tian and Lina Wu from the Central Laboratory, Key Laboratory of Carcinogenesis and Translational Research, Ministry of Education, Beijing Cancer Hospital & Institute for their technical support and assistance. We thank Wendy Bindeman (Mayo Clinic, Immunology Department) for English editing assistance and Chengli Song, from Novogene Company, for her contribution to biology information processing.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.