Introduction

Non-small cell lung cancer (NSCLC) remains a major cause of cancer-related deaths worldwide [1]. Despite numerous advances in our knowledge of NSCLC, development of clinical effective therapies has met with limited success [2]. It underscores the urgent need for refined investigation on molecular mechanisms that provide novel targets for advancing anticancer therapy.

Data from a number of solid tumors support the notion that cancers are maintained by a subpopulation called cancer stem cells (CSCs) [3]. This is also popularly known as the CSCs model which proposes that CSCs are the unique source of all tumor cells at earlier stages of tumorigenesis and, whereas CSCs are responsible for tumor propagation and relapse at advanced stages [4]. Thus, the existence of lung cancer stem cell-like cells (LCSCs) may explain tumorigenesis and propagation of NSCLC. Accumulated findings suggest that targeting LCSCs may be a novel strategy for eradicating NSCLC [5–8]. However, progress in the targeting LCSCs to improve NSCLC treatment has been hindered by lacking understanding of molecular mechanisms that are critical to LCSCs.

SLC34A2, a member of the solute carrier gene family, encodes NaPi2b (the type II Na/Pi co-transporter) [9]. NaPi2b is a multi-transmembrane sodium-dependent phosphate transporter responsible for transcellular inorganic phosphate absorption, maintenance of phosphate homeostasis, and tumorigenesis [10]. It exists in human lung, ovarian, breast, and thyroid cancers [11]. An increased expression of NaPi2b has been reported in ovarian cancer [11], breast cancer [12], and papillary thyroid cancer [13]. Nevertheless, the expression level of SLC34A2 (NaPi2b) in NSCLC patient cohort is an issue of controversy [14–16]. Additionally, the biological function of SLC34A2 and its underlying mechanisms in NSCLC remain unclear. Intriguingly, recent findings have led to an emerging appreciation of the importance of SLC34A2 in maintenance of stem cell phenotype [17] and cell differentiation [18] in cancer. Besides, Ge et al. [12] found that SLC34A2 was necessary for breast cancer stem cell-like cells to induce chemotherapeutic resistance to doxorubicin, which implies that SLC34A2 is likely to play a crucial role in maintaining phenotype of CSCs. However, the biological function and the underlying molecular mechanisms by which SLC34A2 regulates state of LCSCs are unknown.

As for LCSCs, there are some controversies open about the choice of CD166 to enrich for LCSCs. Although some studies concerning CD166 expression in NSCLC deal with its availability to be a LCSCs marker [19], some researchers [20] believe whether CD166 is an appropriate marker for LCSCs must be questioned by evaluating CD166 expression in a large cohort of NSCLC patients. Thus, whether CD166 is able to be the cell-surface marker that identify LCSCs from NSCLC is difficult to pinpoint to date.

Here, we aim to investigate the effects of SLC34A2 in regulating tumorigenicity of CD166+ LCSCs and its underlying mechanisms. Our findings testified that CD166+ cells which were derived from fresh primary NSCLC samples displayed stem cell-like features. Significantly, we found that SLC34A2 expression was enhanced in CD166+ LCSCs. In NSCLC patient cohort, increased SLC34A2 expression correlated with histology, which suggests a potential role of SLC34A2 in CD166+ LCSCs. Furthermore, we demonstrated that SLC34A2 was necessary for self-renewal, proliferation, and tumorigenicity of CD166+ LCSCs. Our study indicated that SLC34A2 regulated Bmi1 to promote tumorigenicity and self-renewal of CD166+ LCSCs through Wnt/β-catenin pathway. Our findings establish a link between SLC34A2 and tumorigenesis, and may provide novel targets for advancing anticancer therapy.

Materials and methods

Sample collection

Fifteen NSCLC primary tissue samples were obtained from a consecutive series of consenting patients according to the approval of Department of Chest Cardiac, the First Hospital of Yulin City from January 2010 to June 2015. All patients were diagnosed with primary NSCLC and did not show other tumor occurrences.

Sphere formation and propagation

Solid NSCLC tissues were finely minced and washed in DMEM/F12 (Gibco Invitrogen, USA). Then they were incubated with Accumax 1X (Innovative Cell Technologies, USA) for 30 min at 37 °C. Single-cell suspension was obtained by filtering digested tissue. Single cells were cultured as the description in Supplemental Information.

Flow cytometry and fluorescence-activated cell sorting (FACS)

Tumor spheres were expanded by enzymatic digestion of spheres with Accumax 1X (Innovative Cell Technologies, USA). Single cells were incubated in staining solution containing 1 % BSA with the specific antibodies at appropriate dilutions. Cells were stained as the description in Supplemental Information.

MTT assay

The MTT assay (Sigma Aldrich, USA) was used to determine relative cell growth every 24 h for cell growth curves. The specific steps were described in Supplemental Information.

In vivo studies of tumorigenicity

All experiments were carried out with female athymic BALB/c nu/nu mice, 3–4 weeks old (HFK Bioscience, China). Mice were maintained at the Animal Core Facility at the First Hospital of Yulin City under specific pathogen-free (SPF) condition. All studies on mice were conducted in accordance with the National Institutes of Health “Guide for the Care and Use of Laboratory Animals.” The protocols of establishment of subcutaneously xenograft models were described in Supplemental Information.

Western-blotting assay were reported in Supplemental Information.

Quantitative real-time RT-PCR analysis (qPCR)

Total RNA for qPCR analysis was extracted using Trizol (Invitrogen, USA), treated with DNase I (Takara, USA) to eliminate contaminating genomic DNA, and reverse-transcribed into cDNA with the Reverse Transcriptase MMLV (Takara, USA). Real-time PCR was performed using a SYBR Green Reagents (Bio-Rad, USA) on the iQ5 Real-Time PCR Detection System (Bio-Rad, USA). Expression of genes with relative to β-actin was determined using the 2 −ΔΔCt method. Primers of these genes were shown in Supplemental Information.

Lentiviral transduction

CD166+ cells were spin-infected with 1 ml of NC-si-RNA (GeneCopoeia, USA), SLC34A2-si-RNA lentiviral knockdowns, or Bmi1-si-RNA lentiviral knockdowns (GeneCopoeia, USA) packbag as detailed in Supplemental Information.

Statistical analysis

SPSS13.0 software was used. Each experiment was performed at least three times. The data were expressed as mean ± SD, and one-way ANOVA and an unpaired Student’s t test were used to determine the significant differences of all the results. Significances are ***P < 0.001, **P < 0.01, and *P < 0.05.

Results

CD166+ cells from NSCLC primary tumor samples display stem cell-like features

In an attempt to identify LCSCs, we analyzed 15 primary tissue samples derived from a consecutive series of NSCLC patients (Fig. 1a). By using the surface marker CD166 alone, FACS analysis showed the presence of a variable fraction of CD166+ cell populations in 15 of 15 tumor samples, varying from 0.2 % to a maximum of 12.1 %. Intriguingly, down-regulation of CD166 expression in patient-derived tumor spheres was observed in adherent cultures grown in 10 % serum-supplemented medium compared with the original cell suspension in serum-free medium, which confirmed the specific expression of CD166 in undifferentiated sphere cells (Fig. 1b). Following sorting CD166+ cells from patient-derived tumor spheres, the self-renewal capacity of CD166+ cells is corroborated by in vitro assays. We tested the CD166+ cells for the ability to form tumor spheres, a widely used in vitro technique for assessing self-renewal capacity. Although both primary CD166+ and CD166− cells remained viable in serum-free medium, only primary CD166+ cells but not CD166− cells were able to form compact self-renewing spheres (Fig. 1c). To evaluate the tumorigenic potential of CD166+ cells, the same numbers of CD166+ and CD166− cells (104 cells) were subcutaneously injected in the flank of nude mice. We found that tumor formation of CD166+ cells was faster and resulted in increased tumor take compared with that observed of CD166− cells (Fig. 1d). To investigate whether lung cancers originating from CD166+ cells possess higher long-term tumorigenic potential compared with the rarely observed tumors originating from the CD166− cells, we performed serial transplantation assays in nude mice of cells isolated from LT1 or LT7 tumor xenografts originally derived from CD166+ or CD166− cell injection. Cells derived from CD166+ tumors were able to generate tumors in primary, secondary, and tertiary transplantation, whereas cells from CD166− tumors lost tumorigenic potential during serial transplantations (Fig. 1d). Therefore, these data indicate that CD166+ cells from NSCLC patient-derived tumor spheres bear stem cell-like features including the ability to self-renew, differentiate, and initiate tumors in vivo.

Fig. 1
figure 1

CD166+ cells from NSCLC primary tumor samples display stem cell-like features. a Clinicopathological characteristics of 15 NSCLC samples and frequency of CD166+ positive cells in these samples. b FACS analysis for CD166 expression in freshly dissociated primary LT7 and corresponding adherent cell culture after three passages in vitro. c Phase-contrast images of tumor spheres seeded with CD166+ and CD166− cells. Scale bar, 100 μm. d In vivo serial transplantation assay. A total of 104 CD166+ cells and CD166− cells, purified from LT1 and LT7, were injected s.c. into nude mice. Derived tumor xenografts were dissociated to single-cell suspension and then serially reinjected in mice (104 cells), generating secondary and then tertiary tumors. Tumor growth curves of primary and tertiary tumors are shown. Note: Columns, mean of three individual experiments; SD,***P < 0.001; SD,**P < 0.01

Full size image

CD166+ LCSCs express high level of SLC34A2

To obtain expression profile of SLC34A2 in CD166+ LCSCs, qPCR was used to compare SLC34A2 expression in CD166+ LCSCs with that in their differentiated adherent progeny and CD166− cells. It showed a significant increase in SLC34A2 expression in CD166+ LCSCs (Fig. 2a). In consistent with SLC34A2 expression profile on mRNA level, we verified that the expression of SLC34A2 significantly enhanced in CD166+ LCSCs (S) compared with that in their differentiated adherent progeny (D) and CD166− cells (A) on protein level (Fig. 2b). Due to the short follow-up period after the fresh clinical specimens were collected, the association between SLC34A2 expression and disease-free survival or overall survival could not be calculated. Nevertheless, qPCR analysis of 15 NSCLC tissue specimens showed that increased SLC34A2 expression correlated with adenocarcinoma more than with squamous cell carcinoma (Fig. 2c). SLC34A2 expression was, however, not significantly associated with age, gender, or advanced disease stage (Fig. 2c). These data suggest a potential role of SLC34A2 in CD166+ LCSCs.

Fig. 2
figure 2

CD166+ LCSCs express high level of SLC34A2. a Relative expression of SLC34A2 in CD166+ LCSCs, their differentiated adherent progeny, and CD166− cells were examined by qPCR. Note: columns, mean of three individual experiments; SD,**P < 0.01. b Relative expression of SLC34A2 in CD166+ LCSCs (S), their differentiated adherent progeny (D), and CD166− cells (A) were examined on protein level. c Clinicopathological correlation of SLC34A2 expression in NSCLC. d CD166+ LCSCs were transduced with lentiviral-based si-RNAs. qPCR analysis was performed for selection of si-SLC34A2 stable expressed cell lines. Note: columns, mean of three individual experiments. e Phase-contrast images of tumor spheres seeded with CD166+ LCSCs and CD166+ LCSCs transduced with si-SLC34A2 or empty vector (si-NC). Scale bar, 100 μm

Full size image

SLC34A2 promotes tumorigenicity and self-renewal capacity of CD166+ LCSCs

High expression of SLC34A2 in CD166+ LCSCs, but not in their differentiated adherent progeny and CD166− cells, suggests that SLC34A2 may drive tumorigenicity and self-renewal capacity of CD166+ LCSCs. To test this hypothesis, we knocked SLC34A2 down in CD166+ LCSCs with lentiviral-based si-RNAs and compared their growth both in vitro and in vivo. Stable transduction with a lentiviral vector containing si-SLC34A2 was selected (Fig. 2d). The ability of CD166+ LCSCs to form spheres and be serially propagated suggested an in vitro self-renewing capacity, whereas the spheres transduced with si-SLC34A2 could not form spheres (Fig. 2e). We found that SLC34A2 was necessary for proliferation in vitro (Fig. 3a). Importantly, tumorigenicity was also significantly reduced upon knockdown of SLC34A2, which suggested that tumorigenesis of CD166+ LCSCs were dependent on SLC34A2 (Figs. 3b). Similarly, the expression of stem cell-associated genes, including Oct4, Sox2, Bmi1, Nanog, and β-catenin, were reduced upon knockdown of SLC34A2 on mRNA level (Fig. 3c). However, on protein level, the expression of Oct4, Sox2, and Nanog were almost unchanged, only the expression of Bmi1 and β-catenin in si-SLC34A2-LCSCs significantly declined compared with those in si-NC-LCSCs (Fig. 4a). These results demonstrate that SLC34A2 promotes tumorigenicity and self-renewal capacity of CD166+ LCSCs and suggest that Bmi1 and β-catenin are likely to be the downstream target of SLC34A2 in CD166+ LCSCs.

Fig. 3
figure 3

SLC34A2 is necessary for proliferation and tumorigenicity of CD166+ LCSCs. a The viability of CD166+ LCSCs and CD166+ LCSCs transduced with si-SLC34A2 or empty vector (si-NC) within 7 days was measured by MTT. Note: columns, mean of three individual experiments; SD,**P < 0.01. b The potential of tumor initiation of CD166+ LCSCs and CD166+ LCSCs transduced with si-SLC34A2 or empty vector (si-NC) by s.c. injection, and representative tumor growth curves of xenografts derived from different cell subpopulations. Note: columns, mean of three individual experiments; SD,**P < 0.01. c qPCR analysis of stemness gene expression in CD166+ LCSCs transduced with si-SLC34A2 or empty vector (si-NC). Note: columns, mean of three individual experiments; SD,**P < 0.01; SD,*P < 0.05

Full size image
Fig. 4
figure 4

Wnt/β-catenin pathway exerts effects on promoting tumorigenesis and self-renewal of CD166+ LCSCs. a Western blot analysis of stemness gene expression in CD166+ LCSCs transduced with si-SLC34A2 or empty vector (si-NC). b With or without the presence of Wnt3a, phase-contrast images of tumor spheres seeded with CD166− cells. Scale bar, 100 μm. c With or without the presence of XAV939, phase-contrast images of tumor spheres seeded with CD166+ LCSCs. Scale bar, 100 μm. d With or without the presence of Wnt3a, the viability of CD166− cells within 7 days was measured by MTT. With or without the presence of XAV939, the viability of CD166+ LCSCs within 7 days was measured by MTT. Note: columns, mean of three individual experiments; SD,**P < 0.01. e Relative expression of Bmi1 in CD166+ LCSCs with or without the presence of XAV939 and relative expression of Bmi1 in CD166− cells with or without the presence of Wnt3a were examined by qPCR. Note: columns, mean of three individual experiments; SD,**P < 0.01

Full size image

SLC34A2 regulates Bmi1 to promote tumorigenicity and self-renewal capacity of CD166+ LCSCs through Wnt/β-catenin pathway

In an effort to determine whether Wnt/β-catenin pathway was potential downstream target of SLC34A2 to promote tumorigenicity and self-renewal capacity of CD166+ LCSCs, XAV939 (the inhibitor of Wnt/β-catenin pathway) and Wnt3a (the inducer of Wnt/β-catenin pathway) were used. We found that the ability of sphere formation in CD166− cells was enhanced with the presence of Wnt3a (Fig. 4b), whereas with the presence of XAV939, the ability of sphere formation in CD166+ LCSCs was significantly reduced (Fig. 4c). Similarly, proliferation of CD166+ LCSCs in vitro declined with the presence of XAV939, while proliferation of CD166- cells was enhanced with the presence of Wnt3a (Fig. 4d). Intriguingly, the expression of Bmi1 in CD166− cells was increased with the presence of Wnt3a, whereas with the presence of XAV939, Bmi1 expression in CD166+ LCSCs was reduced (Fig. 4e). These data indicate that Wnt/β-catenin pathway exerts effects on promoting tumorigenicity and self-renewal capacity of CD166+ LCSCs and imply that Bmi1 may be the downstream target of Wnt/β-catenin pathway.

To determine whether Bmi1 could be exploited by SLC34A2 to promote tumorigenicity and self-renewal capacity of CD166+ LCSCs, we knocked Bmi1 down in CD166+ LCSCs with lentiviral-based si-RNAs. Stable transduction with a lentiviral vector containing si-Bmi1 in CD166+ LCSCs was selected (Fig. 5a, b). We found that tumorigenicity in vivo and proliferation in vitro were both significantly reduced upon knockdown of Bmi1 (Fig. 5c, d), suggesting that Bmi1 was necessary for tumorigenicity and proliferation of CD166+ LCSCs and Bmi1 was the downstream target of SLC34A2.

Fig. 5
figure 5

Bmi1 is necessary for tumorigenicity and proliferation of CD166+ LCSCs. a Phase-contrast and fluorescence images of CD166+ LCSCs which were transduced with lentiviral-based si-Bmi1. Scale bar, 100 mm. b qPCR analysis was performed for selection of si-Bmi1 stable expressed cell lines. Note: columns, mean of three individual experiments. c The potential of tumor initiation of CD166+ LCSCs and CD166+ LCSCs transduced with si-Bmi1 or empty vector (si-NC) by s.c. injection, and representative tumor growth curves of xenografts derived from different cell subpopulations. Note: columns, mean of three individual experiments; SD,**P < 0.01. d The viability of CD166+ LCSCs and CD166+ LCSCs transduced with si-Bmi1 or empty vector (si-NC) within 7 days was measured by MTT. Note: columns, mean of three individual experiments; SD,**P < 0.01

Full size image

Furthermore, we characterized the relationship among SLC34A2, Wnt/β-catenin pathway, and Bmi1. We found that the expression of SLC34A2, β-catenin, Dvl2, and Bmi1 in CD166+ LCSCs was reduced upon knockdown of SLC34A2. However, with the presence of Wnt3a, the expression of β-catenin, Dvl2, and Bmi1 in si-SLC34A2-LCSCs was upregulated, which suggested that both of Wnt/β-catenin pathway and Bmi1 were downstream targets of SLC34A2 (Fig. 6). With the presence of XAV939, the expression of β-catenin, Dvl2, and Bmi1 in CD166+ LCSCs were significantly reduced, while SLC34A2 expression was unchanged (Fig. 6). Additionally, the expression of Bmi1 in CD166+ LCSCs was reduced upon knockdown of Bmi1, whereas the expression of SLC34A2, β-catenin, and Dvl2 were unchanged, which implied that Bmi1 were downstream target of Wnt/β-catenin pathway (Fig. 6). Taken together, these results suggest SLC34A2 regulates Bmi1 to promote tumorigenicity of CD166+ LCSCs through Wnt/β-catenin pathway.

Fig. 6
figure 6

SLC34A2 regulates Bmi1 to promote tumorigenesis and self-renewal of CD166+ LCSCs through Wnt/β-catenin pathway. With or without the presence of Wnt3a or XAV939, western blot analysis was performed to determine the expression of SLC34A2, β-catenin, Dvl2, and Bmi1 in CD166+ LCSCs and CD166+ LCSCs transduced with si-SLC34A2 or si-Bmi1. β-actin was used as a normalization control

Full size image

Discussion

The identification of distinct phenotypic and molecular signature associated with tumorigenicity of LCSCs would be instrumental for developing therapeutic targeting strategies in NSCLC patients. Our work sheds new light on the nature of the LCSCs state and the role of SLC34A2 in tumorigenicity of LCSCs.

In this study, our data showed that CD166 served as an inert surface marker. Following culturing cells from 15 cases of NSCLC primary fresh tumor samples in serum-free medium for 3–5 weeks, a few tumor cells grew to form spheres. CD166+ cells isolated from primary lung tumors showed higher tumorigenic and self-renewal potential than their CD166− counterparts. Additionally, CD166+ spheres readily generated differentiated cells closely resembling the main adherent cells of the original tumor in 10 % serum cultivation. Our data indicate that CD166+ cells from NSCLC patient-derived tumor spheres bear stem cell-like features. On clinic, we analyzed 15 primary tissue samples derived from a consecutive series of NSCLC patients. The expression of CD166 was not significantly associated with age, gender, or advanced disease stage. This is consistent with what Tachezy et al. [20] found.

To date, the molecular basis of SLC34A2 in LCSCs is unknown and underlying mechanisms in LCSCs remain obscure. Here, a significant overexpression of SLC34A2 distinguished CD166+ LCSCs from their differentiated progeny and CD166− cells. qPCR analysis of 15 NSCLC tissue specimens showed that increased SLC34A2 expression correlated with adenocarcinoma more than with squamous cell carcinoma. These data suggest a potential role of SLC34A2 in CD166+ LCSCs, but more NSCLC tissue specimens, as well as follow-up information, should be enrolled for investigation of SLC34A2 expression prognostic value. On the other hand, functional studies found that SLC34A2 was required for self-renewal and proliferation in vitro, and tumorigenicity in vivo. The knockdown of SLC34A2 in CD166+ LCSCs reduced proliferation, self-renewal capacity, the expression of stem cell-associated genes, and tumorigenicity in vivo when compared with control cells transduced with empty vector alone.

In light of the a significant decrease in the expression of Bmi1 and β-catenin upon knockdown of SLC34A2, we hypothesized that Bmi1 and β-catenin were likely to be the downstream target of SLC34A2 in CD166+ LCSCs. By testing the function of Wnt/β-catenin pathway and Bmi1 with Wnt3a, XAV939, and lentiviral-based si-Bmi1, we demonstrated that Wnt/β-catenin pathway and Bmi1 were both necessary for tumorigenic and self-renewal potential of CD166+ LCSCs. Furthermore, we indicated that Bmi1 was the downstream target of Wnt/β-catenin pathway. SLC34A2 regulated Bmi1 to promote tumorigenic and self-renewal potential of CD166+ LCSCs through Wnt/β-catenin pathway.

In summary, our findings testified that CD166+ cells which were derived from fresh primary NSCLC samples displayed stem cell-like features. Significantly, we found that SLC34A2 expression was enhanced in CD166+ LCSCs. In NSCLC patient cohort, increased SLC34A2 expression correlated with histology, which suggests a potential role of SLC34A2 in CD166+ LCSCs. Furthermore, we demonstrated that SLC34A2 was necessary for self-renewal capacity, proliferation, and tumorigenicity of CD166+ LCSCs. Our study indicated that SLC34A2 regulated Bmi1 to promote tumorigenic and self-renewal potential of CD166+ LCSCs through Wnt/β-catenin pathway. In this study, the characterization of molecular basis of SLC34A2 in CD166+ LCSCs not only allows for better understanding of the mechanisms regulating this specific population of NSCLC cells but also provides insight into the gradual improvement of more effective cancer therapies against this disease.